ABSTRACT
Background
Current dietary guidance recommends limiting intakes of saturated fats, but most fails to consider that saturated fats from different food sources may have different health effects.
Objectives
We aimed to evaluate the associations of saturated fats from dairy and nondairy sources with measures of body fat, inflammatory biomarkers, lipid concentrations, and lipid particle sizes and concentrations.
Methods
The Framingham Offspring Study is a prospective cohort study. Participants (n = 2391) ≥30 y of age who had dietary records and data on the outcomes of interest were included.
Results
Among females, those in the highest quintile (compared with the lowest) of dairy-derived saturated fat had lower multivariable-adjusted levels of body fat [BMI (in kg/m2): 26.2 compared with 27.8, P < 0.01; and percentage fat mass: 36.7% compared with 38.0%, P = 0.09) and larger LDL particle sizes. Nondairy saturated fat in females was inversely associated with the triglyceride (TG):HDL ratio (P = 0.03). Among males, intakes of dairy-derived saturated fats were inversely associated with C-reactive protein (P < 0.01), fibrinogen (P < 0.01), TGs (P < 0.01), and the TG:HDL ratio (P < 0.01). HDL cholesterol was 2.8 mg/dL (P = 0.04) higher among males in the highest (compared with the lowest) quintile of saturated fat from dairy sources. Males with the highest intakes of dairy-derived saturated fats had larger HDL and LDL particle sizes (P < 0.01 for both), a higher HDL particle concentration (P < 0.01), and a lower VLDL particle concentration (P < 0.01). There were no statistically significant adverse effects of saturated fats from nondairy sources on any of these outcomes in either males or females.
Conclusions
Males with higher intakes of dairy-derived saturated fats had a less atherogenic profile than males with lower intakes of these fats. These effects were weaker in females. Nondairy saturated fats were not associated with these cardiometabolic outcomes.
Keywords: diet, saturated fat, dairy, cardiometabolic risk, lipids, body fat, inflammation, lipid particles
See corresponding editorial on page 1466.
Introduction
Cardiovascular disease (CVD) is the leading cause of death in America and dyslipidemia is a major contributor to its development. The diet-heart hypothesis first emerged >60 y ago and suggested that saturated fat was a critical preventable cause of death and disability from CVD (1) as well as a determinant of other cardiometabolic disorders such as obesity and dyslipidemia. However, more recent evidence suggests that saturated fat may not be a strong causal determinant of the development of CVD (2–4). Still, the current US Dietary Guidelines recommend limiting saturated fat intake and replacing saturated fats such as those in full-fat dairy and other animal fats with unsaturated fatty acids (5). These guidelines, however, fail to consider that different saturated fat–containing foods may have very different health effects.
Dairy foods are important contributors to the intake of saturated fat in the American diet and although total saturated fat may be positively associated with dyslipidemia in some studies, a 2020 review found that whole-fat dairy foods are not associated with increased risk of CVD or diabetes (6). Other studies have also shown that whole-fat dairy intakes can reduce the risk of diabetes (2), and dairy fat has also been shown to have beneficial effects on lipids, such as triglyceride (TG) concentrations, in others (7). A meta-analysis concluded that although lauric acid (12:0; found in butter) raised LDL and total cholesterol, it also increased HDL-cholesterol concentrations, as well as the total:HDL cholesterol ratio (8). The 2015 Dietary Guidelines Advisory Committee acknowledged the need for more systematic data on the effects of different food sources of saturated fat on lipids and other metabolic outcomes.
Dietary saturated fats are also associated with the intake of energy-dense foods and, as a result, it has been suggested that they may also promote excess weight gain. The aims of this study were to compare the effects of saturated fat from dairy and nondairy sources on measures of body fat, biomarkers of inflammation, lipid concentrations, and lipid particle sizes and numbers, separately for males and females. Specifically, we examined effects on BMI (in kg/m2), percentage body fat, waist-to-height ratio, C-reactive protein (CRP) concentrations, fibrinogen, LDL cholesterol, HDL cholesterol, TG, and the TG:HDL ratio, as well as particle sizes and concentrations for LDL, HDL, and VLDL, in a prospective community-based cohort study.
Methods
The Framingham Offspring Study is a prospective cohort that began in 1971 by enrolling adult offspring (and their spouses) of the original Framingham Heart Study participants. The current analyses were designed to evaluate the effects of saturated fat from dairy and nondairy sources on body fat, inflammation, and lipids. We included participants who were 30 y of age or older and who provided dietary data during the third and fifth examination cycles. From that point forward, participants were followed up at clinic examination visits scheduled every 4 y. All participants provided informed consent and all of the methods used met the ethical standards set forth by the Boston University School of Medicine. The original data collection and these analyses were approved by the Boston University School of Medicine Institutional Review Board.
Figure 1 provides a flowchart outlining the selection of the study populations for each of the analyses. Of the 5124 original Framingham Offspring participants, 461 died before examination visit 5, 65 were too young at the time of the dietary assessment, and 205 failed to provide follow-up data during the study period. Another 1493 were missing dietary record data and 347 were excluded owing to extreme amounts of intake as described in Figure 1. Finally, 21 participants with BMI < 18.5, 5 who were missing baseline weight, and 136 with prevalent cancer were also excluded, leaving 2391 participants remaining for the analyses. Supplemental Table 1 provides a comparison of characteristics of subjects who were included and those who were excluded. Individuals missing outcome data for body fat, inflammatory biomarkers, blood lipids, or lipid particles were then excluded from outcome-specific analyses, respectively. This left 2222, 2021, 2066, and 1874 participants for the analysis of body fat, inflammation, lipids, and lipid particles, respectively.
FIGURE 1.
Inclusion and exclusion criteria. Extreme energy intakes included <1200 or >4000 kcal/d for males and <1000 or >3500 kcal/d for females. Exclusions for extreme alcohol intake in both males and females included those with >20% of daily kilocalories from alcohol. Extreme dietary fat values included extreme intakes of individual fat-containing foods (e.g., >8 oz/d of red meat, poultry, or fish; >14 oz of nuts, seeds, soy) or extreme intakes of saturated or unsaturated dietary fats (e.g., <4% or >20% of energy from SFA or MUFA or <2% or ≥17% of energy from PUFA). 1 oz = 28.35 g.
Exposure variables: dietary intakes of saturated fat from dairy and nondairy sources
Diet was assessed using 2 sets of 3-d diet records collected during examination cycles 3 and 5. Participants were instructed in the completion of the diet records by a trained nutritionist. The instruction included the use of 2-dimensional food models to assist in estimating portion sizes and the provision of detailed data on brands and recipes. The dietary records were analyzed for nutrient content by a separate blinded nutritionist using the Nutrition Data System for Research (NDSR) of the University of Minnesota (9). In the data cleaning process, extreme values for individual foods and nutrients were used to identify errors in data entry and reporting. When extremes were noted, we returned to the original dietary records to correct any data entry errors and then marked individual days for deletion when they appeared to be unfeasible or when they did not reflect usual intake (e.g., subject was ill).
Given the time period in which the Framingham dietary data were collected, we used data from a contemporaneous Microsoft disk operating system (DOS) version of NDSR to derive daily saturated fat intake from dairy foods. NDSR disaggregates many composite foods (e.g., pancakes) into component ingredients (e.g., flour, milk, egg). The majority of dairy intake was derived from individual dairy foods (i.e., milk, yogurt, cheeses) that came from the disaggregated information in NDSR. For these, we were able to assign all saturated fat from individual dairy foods to dairy-derived saturated fat. We also searched for foods not counted as dairy equivalents but which provided discretionary fat from dairy sources (e.g., butter, cream). Some foods, however, are not disaggregated in NDSR, making it impossible to derive dairy fat directly. In this situation, we identified matching food codes in the MyPyramid servings database and used the dairy servings assigned to that food to derive the associated saturated fat content. In this way, we were able to sum the saturated fat intake per day from individual dairy foods from the contemporaneous DOS version of NDSR with the saturated fat intake from composite foods containing dairy from the MyPyramid servings database. Nondairy saturated fat was derived as total saturated fat minus dairy-related saturated fat. Finally, mean intakes of saturated fats from dairy and nondairy sources were calculated using all available days of dietary data. Saturated fats from both dairy and nondairy sources were adjusted for body weight using the residual method. The residuals from the linear regression models were then added to the overall median intakes for their respective nutrients.
Outcome variables: body fat, inflammation, lipid concentrations, and lipid particles
At each examination visit, height (to the nearest 0.25 inch, which is equal to 0.64 cm) and weight (to the nearest 0.25 pound,which is equal to 0.11 kg) were measured using a standard counterbalance scale with a stadiometer. To reduce random measurement error and the effect of height loss after the age of 60 y, we used the average adult height for each subject up to the age of 60 y combined with exam-specific weight (in kg) to calculate exam-specific BMI. A bioelectrical impedance analyzer (BIA) was used at each examination to measure body fat; a standard tetrapolar technique was followed and percentage body fat was estimated (BIA-101, RJL Systems) using a previously described validated method (10).
Routine blood samples were collected from all subjects after a 12-h fast and plasma aliquots were stored at −70°C before analysis. For these analyses, blood measures were taken from the visit ∼4 y after the final baseline dietary assessment and used to determine high-sensitivity CRP concentrations, fibrinogen, and lipids. High-sensitivity CRP was measured using a standard assay (Dade Behring) and fibrinogen was measured using the Clauss method (Diagnostica Stago reagents). Lipid concentrations were determined following standardized protocols in the Framingham laboratory, which is a standardized lipoprotein cholesterol laboratory following guidelines from the CDC. HDL and TG concentrations were determined using standard enzymatic methods whereas LDL was estimated using the Friedewald equation (11). Lipid particle sizes were evaluated using an NMR spectroscopic assay (LipoScience) (12). Mean particle sizes were derived by taking the sum of the diameters within each subclass and multiplying by its relative mass percentage shown by the amplitude of the NMR signal. All biochemical and NMR analyses were carried out in a blinded fashion.
Potential confounding variables
A large number of potential confounders were assessed routinely in Framingham. Self-reported education level was collected at examination visit 2 and used as a marker of socioeconomic status. Cigarette smoking was assessed by self-report at every examination visit. Pack-years of cigarettes were calculated from age 18 y to the time of the baseline exam. Alcohol intake was estimated as mean intake per day of beer, wine, and spirits (converted to grams of alcohol) during the dietary assessment period (examinations 3–5). Physical activity was assessed at examination 4 by asking each subject to report the number of hours per day spent in sleep, sedentary, light, moderate, and vigorous activities in a 24-h period. As previously described, moderate and vigorous physical activities were assigned a score using an appropriate weight based on the estimated energy expenditure (oxygen consumption) required for that level of activity (13). A physical activity index was derived as the sum of the weighted moderate and vigorous activity scores. Various dietary factors were considered as potential confounders including intakes of other dietary fats, carbohydrates, fruits and vegetables as well as nonstarchy vegetables, whole grains, and fiber. Scores on the 2015 Healthy Eating Index (HEI-2015) were used as an indicator of diet quality (https://epi.grants.cancer.gov/hei/sas-code.html).
Statistical analysis
Dietary intakes of weight-adjusted saturated fats from dairy and nondairy sources were averaged over the 8 y of dietary records. We then categorized mean intakes of saturated fat from dairy and nondairy sources into quintiles. We then compared the adjusted mean levels of body fat, biomarkers of inflammation, and blood lipids 4 y later. ANCOVA models were used to adjust for confounding by those factors that altered the mean differences in each outcome by ≥5%. The final models included age, sex, weight-adjusted intakes of carbohydrates and MUFAs, scores on the HEI-2015, and baseline BMI (for inflammatory biomarkers and lipid-related outcomes). Energy intake was not a confounder in these analyses.
Results
Participants on average were in their mid-50s at the end of the baseline dietary assessment. Mean ± SD overall intake of saturated fat from dairy sources was 10.2 ± 6.8 g for males and 9.0 ± 5.3 g for females, whereas that from nondairy sources was 17.6 ± 6.6 g for males and 14.3 ± 5.4 g for females. Table 1 shows the characteristics of the participants according to intakes of dairy-derived saturated fat, in quintiles. The ranges of intake in the quintiles in both males and females were as follows: 0 to <4.5, 4.5 to <7.1, 7.1 to <9.9, 9.9 to <13.7, and 13.7–40.4 g/d. In Table 1, both males and females in the higher quintiles of saturated fat intake from dairy sources also had higher intakes of energy, higher weight-adjusted macronutrient intakes, and higher intakes of foods from most protein-related sources except poultry and fish. The same was true in association with saturated fat from nondairy sources (Supplemental Table 2). Saturated fat, regardless of source, was inversely associated with scores on the HEI-2015.
TABLE 1.
Characteristics of males and females at baseline according to quintiles of dairy-derived saturated fat intakes1
| Saturated fat from dairy sources | Q1 (0.0 to <4.5 g/d) | Q2 (4.5 to <7.1 g/d) | Q3 (7.1 to <9.9 g/d) | Q4 (9.9 to <13.7 g/d) | Q5 (13.7–40.4 g/d) | P-trend2 |
|---|---|---|---|---|---|---|
| Males | ||||||
| n | 216 | 188 | 192 | 210 | 285 | |
| Age, y | 58.8 ± 8.7 | 57.0 ± 10.0 | 57.2 ± 9.8 | 54.8 ± 9.8 | 54.0 ± 10.0 | <0.01 |
| BMI, kg/m2 | 28.8 ± 4.4 | 27.8 ± 3.6 | 27.5 ± 3.8 | 27.8 ± 3.9 | 27.6 ± 3.8 | <0.01 |
| Physical activity, metabolic eq/d | 16.2 ± 8.6 | 14.8 ± 8.3 | 14.4 ± 8.5 | 14.9 ± 8.7 | 15.2 ± 9.4 | 0.35 |
| Cigarette smoking, pack-years | 21.1 ± 22.4 | 19.1 ± 21.7 | 19.1 ± 23.8 | 15.4 ± 21.2 | 19.9 ± 23.9 | 0.29 |
| Alcohol, g/d | 14.4 ± 17.0 | 12.4 ± 14.5 | 11.0 ± 13.4 | 12.1 ± 15.5 | 13.7 ± 16.9 | 0.76 |
| Dietary intake | ||||||
| Energy, kcal/d | 1907 ± 440 | 1993 ± 468 | 2112 ± 455 | 2286 ± 440 | 2511 ± 470 | <0.01 |
| Meats/processed meats, oz-eq/d | 2.9 ± 1.9 | 2.8 ± 1.7 | 3.0 ± 1.6 | 3.2 ± 1.6 | 3.3 ± 1.7 | <0.01 |
| Poultry, oz-eq/d | 1.9 ± 1.6 | 1.6 ± 1.4 | 1.5 ± 1.2 | 1.2 ± 1.1 | 1.4 ± 1.4 | <0.01 |
| Fish, oz-eq/d | 1.6 ± 1.5 | 1.4 ± 1.4 | 1.3 ± 1.2 | 1.2 ± 1.4 | 1.3 ± 1.4 | 0.01 |
| Dairy, cup-eq/d | 0.8 ± 0.6 | 1.2 ± 0.9 | 1.4 ± 0.7 | 1.7 ± 0.8 | 2.2 ± 1.0 | <0.01 |
| Fruit, nonstarchy veg, cup-eq/d | 3.0 ± 1.7 | 2.7 ± 1.7 | 2.5 ± 1.3 | 2.6 ± 1.5 | 2.4 ± 1.3 | <0.01 |
| HEI-2015 score | 61.4 ± 10.8 | 57.9 ± 10.5 | 55.7 ± 10.7 | 53.0 ± 10.0 | 49.0 ± 8.9 | <0.01 |
| Nutrients, g/d | ||||||
| Carbohydrates | 223.6 ± 69.0 | 236.7 ± 78.1 | 245.0 ± 69.7 | 261.7 ± 65.6 | 273.6 ± 66.9 | <0.01 |
| Protein | 84.7 ± 19.5 | 84.6 ± 22.4 | 87.2 ± 19.8 | 90.3 ± 19.3 | 98.2 ± 21.3 | <0.01 |
| Total fat | 66.8 ± 19.5 | 72.3 ± 21.9 | 81.6 ± 22.1 | 91.5 ± 23.2 | 106.9 ± 23.6 | <0.01 |
| Saturated fat | 19.4 ± 5.9 | 23.0 ± 6.5 | 27.1 ± 7.0 | 32.0 ± 7.0 | 40.9 ± 8.6 | <0.01 |
| Dairy SFA | 3.2 ± 1.4 | 6.1 ± 1.0 | 8.8 ± 1.0 | 12.2 ± 1.3 | 19.9 ± 5.2 | <0.01 |
| Nondairy SFA | 16.3 ± 5.8 | 16.9 ± 6.2 | 18.3 ± 6.7 | 19.8 ± 6.8 | 21.0 ± 6.7 | <0.01 |
| Nutrients, g/d (wt-adjusted) | ||||||
| Carbohydrates | 213.6 ± 72.0 | 230.2 ± 79.4 | 238.8 ± 71.3 | 254.0 ± 67.0 | 265.7 ± 67.0 | <0.01 |
| Protein | 79.1 ± 21.1 | 81.0 ± 21.8 | 83.8 ± 20.1 | 86.0 ± 19.4 | 93.8 ± 21.4 | <0.01 |
| Total fat | 61.2 ± 20.9 | 68.6 ± 20.6 | 78.1 ± 22.1 | 87.2 ± 23.0 | 102.4 ± 23.1 | <0.01 |
| Saturated fat | 17.5 ± 6.3 | 21.8 ± 6.0 | 25.9 ± 6.9 | 30.5 ± 6.8 | 39.4 ± 8.4 | <0.01 |
| Dairy SFA | 2.6 ± 1.4 | 5.7 ± 0.8 | 8.4 ± 0.8 | 11.7 ± 1.1 | 19.4 ± 5.2 | <0.01 |
| Nondairy SFA | 14.9 ± 6.1 | 16.0 ± 5.9 | 17.4 ± 6.7 | 18.7 ± 6.7 | 19.9 ± 6.6 | <0.01 |
| Females | ||||||
| n | 254 | 298 | 280 | 276 | 192 | |
| Age, y | 58.4 ± 8.9 | 56.2 ± 9.7 | 55.3 ± 9.6 | 54.0 ± 9.2 | 52.7 ± 9.4 | <0.01 |
| BMI, kg/m2 | 28.2 ± 5.9 | 26.2 ± 4.9 | 26.2 ± 5.1 | 25.7 ± 4.7 | 25.1 ± 4.6 | <0.01 |
| Physical activity, metabolic eq/d | 14.1 ± 6.9 | 14.5 ± 7.5 | 14.3 ± 7.8 | 14.0 ± 7.3 | 14.6 ± 7.5 | 0.83 |
| Cigarette smoking, pack-years | 13.4 ± 18.4 | 13.0 ± 18.6 | 12.7 ± 19.4 | 11.6 ± 18.3 | 12.7 ± 19.3 | 0.41 |
| Alcohol, g/d | 5.3 ± 8.1 | 5.3 ± 8.2 | 6.7 ± 9.8 | 7.0 ± 9.8 | 6.9 ± 10.1 | <0.01 |
| Dietary intake | ||||||
| Energy, kcal/d | 1443 ± 345 | 1510 ± 282 | 1630 ± 355 | 1775 ± 373 | 1974 ± 428 | <0.01 |
| Meats/processed meats, oz-eq/d | 1.8 ± 1.2 | 1.9 ± 1.3 | 2.0 ± 1.2 | 2.1 ± 1.4 | 2.1 ± 1.4 | <0.01 |
| Poultry, oz-eq/d | 1.5 ± 1.3 | 1.3 ± 1.1 | 1.3 ± 1.1 | 1.2 ± 1.2 | 1.0 ± 1.0 | <0.01 |
| Fish, oz-eq/d | 1.3 ± 1.2 | 1.2 ± 1.2 | 1.1 ± 1.1 | 1.1 ± 1.0 | 1.1 ± 1.1 | 0.01 |
| Dairy, cup-eq/d | 0.8 ± 0.5 | 1.1 ± 0.9 | 1.2 ± 0.6 | 1.4 ± 0.7 | 1.8 ± 0.8 | <0.01 |
| Fruit, nonstarchy veg, cup-eq/d | 2.7 ± 1.4 | 2.5 ± 1.1 | 2.4 ± 1.2 | 2.5 ± 1.3 | 2.4 ± 1.3 | 0.06 |
| HEI-2015 score | 63.8 ± 11.5 | 61.4 ± 11.4 | 57.4 ± 10.8 | 55.0 ± 10.7 | 51.3 ± 10.1 | <0.01 |
| Nutrients, g/d | ||||||
| Carbohydrates | 177.7 ± 56.5 | 181.4 ± 48.3 | 191.0 ± 53.1 | 203.9 ± 53.6 | 221.8 ± 63.9 | <0.01 |
| Protein | 65.2 ± 15.6 | 66.7 ± 14.7 | 68.6 ± 16.9 | 72.3 ± 17.9 | 76.4 ± 18.8 | <0.01 |
| Total fat | 51.1 ± 17.0 | 56.1 ± 15.4 | 63.4 ± 17.4 | 72.3 ± 19.3 | 85.1 ± 21.8 | <0.01 |
| Saturated fat | 15.0 ± 5.4 | 17.7 ± 4.9 | 21.2 ± 5.2 | 25.6 ± 6.0 | 33.0 ± 7.7 | <0.01 |
| Dairy SFA | 2.8 ± 1.3 | 5.4 ± 0.9 | 7.9 ± 1.1 | 11.2 ± 1.3 | 18.1 ± 4.5 | <0.01 |
| Nondairy SFA | 12.2 ± 5.1 | 12.3 ± 4.7 | 13.3 ± 4.9 | 14.5 ± 5.7 | 14.9 ± 5.7 | <0.01 |
| Nutrients, g/d (wt-adjusted) | ||||||
| Carbohydrates | 179.9 ± 58.4 | 188.1 ± 49.7 | 198.2 ± 54.2 | 211.8 ± 54.1 | 230.4 ± 63.3 | <0.01 |
| Protein | 66.5 ± 15.8 | 70.4 ± 15.7 | 72.6 ± 16.5 | 76.6 ± 18.1 | 81.1 ± 17.8 | <0.01 |
| Total fat | 52.3 ± 17.1 | 59.9 ± 15.8 | 67.5 ± 17.7 | 76.7 ± 19.1 | 89.9 ± 20.8 | <0.01 |
| Saturated fat | 15.5 ± 5.3 | 19.0 ± 5.0 | 22.7 ± 5.1 | 27.2 ± 5.9 | 34.8 ± 7.3 | <0.01 |
| Dairy SFA | 2.9 ± 1.1 | 5.8 ± 0.7 | 8.4 ± 0.8 | 11.6 ± 1.1 | 18.6 ± 4.4 | <0.01 |
| Nondairy SFA | 12.5 ± 5.1 | 13.3 ± 4.8 | 14.3 ± 5.0 | 15.6 ± 5.7 | 16.2 ± 5.5 | <0.01 |
Values are mean ± SD unless otherwise indicated. eq, equivalents; HEI, Healthy Eating Index; Q, quintile.
P-trend was derived from a simple linear regression model.
We examined the trends in adjusted measures of body fat after an average of 4 y of follow-up across quintiles of dairy and nondairy saturated fat intakes among males and females in Table 2. Females in the highest quintile of saturated fat from dairy sources had a mean BMI at the end of follow-up of 26.2 compared with 27.8 for those in the lowest quintile. Similarly, females with the lowest intakes of dairy fat had the highest percentage body fat. There was no association between quintiles of nondairy-derived saturated fat and measures of body fat in males.
TABLE 2.
Sex-specific associations between quintiles of saturated fat from dairy and nondairy sources and mean levels of body fat and inflammatory markers after 4 y of follow-up1
| Type of fat | Range | n | BMI,2 kg/m2 | Fat mass,2 % | Waist:height ratio2 | n | Log(CRP)3 | Log (fibrinogen)3 |
|---|---|---|---|---|---|---|---|---|
| Males | ||||||||
| Dairy-derived saturated fat intake | ||||||||
| Q1 | 0.0 to <4.5 g/d | 193 | 28.1 ± 0.26 | 28.7 ± 0.48 | 0.58 ± 0.004 | 178 | 1.18 ± 0.04 | 5.82 ± 0.01 |
| Q2 | 4.5 to <7.1 g/d | 169 | 27.3 ± 0.27 | 27.6 ± 0.48 | 0.57 ± 0.004 | 156 | 1.06 ± 0.04 | 5.80 ± 0.01 |
| Q3 | 7.1 to <9.9 g/d | 168 | 27.1 ± 0.26 | 27.4 ± 0.47 | 0.56 ± 0.004 | 163 | 1.08 ± 0.04 | 5.76 ± 0.01 |
| Q4 | 9.9 to <13.7 g/d | 183 | 27.3 ± 0.25 | 27.3 ± 0.46 | 0.57 ± 0.004 | 182 | 1.04 ± 0.04 | 5.75 ± 0.01 |
| Q5 | 13.7–40.4 g/d | 247 | 27.5 ± 0.24 | 27.8 ± 0.43 | 0.57 ± 0.004 | 252 | 1.00 ± 0.04 | 5.75 ± 0.01 |
| P-trend4 | 0.21 | 0.23 | 0.18 | 0.01 | <0.01 | |||
| Nondairy saturated fat intake | ||||||||
| Q1 | 0.0 to <10.5 g/d | 132 | 27.7 ± 0.40 | 28.0 ± 0.71 | 0.57 ± 0.006 | 138 | 0.99 ± 0.06 | 5.74 ± 0.02 |
| Q2 | 10.5 to <13.4 g/d | 139 | 27.7 ± 0.32 | 27.5 ± 0.57 | 0.57 ± 0.005 | 128 | 1.10 ± 0.05 | 5.77 ± 0.02 |
| Q3 | 13.4 to <16.4 g/d | 180 | 27.3 ± 0.26 | 27.8 ± 0.47 | 0.57 ± 0.004 | 169 | 1.01 ± 0.04 | 5.76 ± 0.01 |
| Q4 | 16.4 to <20.6 g/d | 218 | 27.1 ± 0.23 | 27.6 ± 0.42 | 0.56 ± 0.004 | 206 | 1.10 ± 0.04 | 5.80 ± 0.01 |
| Q5 | 20.6–41.4 g/d | 291 | 27.7 ± 0.29 | 28.0 ± 0.52 | 0.57 ± 0.004 | 290 | 1.10 ± 0.05 | 5.79 ± 0.01 |
| P-trend4 | 0.50 | 0.98 | 0.78 | 0.34 | 0.06 | |||
| Females | ||||||||
| Dairy-derived saturated fat intake | ||||||||
| Q1 | 0.0 to <4.5 g/d | 242 | 27.8 ± 0.34 | 38.0 ± 0.46 | 0.59 ± 0.006 | 210 | 1.16 ± 0.05 | 5.80 ± 0.01 |
| Q2 | 4.5 to <7.1 g/d | 294 | 26.4 ± 0.29 | 36.7 ± 0.40 | 0.57 ± 0.005 | 249 | 1.27 ± 0.04 | 5.81 ± 0.01 |
| Q3 | 7.1 to <9.9 g/d | 270 | 26.4 ± 0.30 | 36.4 ± 0.41 | 0.57 ± 0.005 | 234 | 1.20 ± 0.04 | 5.80 ± 0.01 |
| Q4 | 9.9 to <13.7 g/d | 269 | 26.4 ± 0.30 | 36.6 ± 0.42 | 0.57 ± 0.005 | 231 | 1.17 ± 0.04 | 5.79 ± 0.01 |
| Q5 | 13.7–40.4 g/d | 187 | 26.2 ± 0.39 | 36.7 ± 0.53 | 0.58 ± 0.007 | 166 | 1.19 ± 0.05 | 5.78 ± 0.01 |
| P-trend4 | 0.01 | 0.09 | 0.09 | 0.73 | 0.15 | |||
| Nondairy saturated fat intake | ||||||||
| Q1 | 0.0 to <10.5 g/d | 308 | 27.5 ± 0.38 | 37.9 ± 0.53 | 0.59 ± 0.007 | 260 | 1.21 ± 0.05 | 5.78 ± 0.01 |
| Q2 | 10.5 to <13.4 g/d | 308 | 26.1 ± 0.29 | 36.4 ± 0.40 | 0.57 ± 0.005 | 270 | 1.21 ± 0.04 | 5.79 ± 0.01 |
| Q3 | 13.4 to <16.4 g/d | 265 | 26.3 ± 0.30 | 36.7 ± 0.41 | 0.58 ± 0.005 | 222 | 1.22 ± 0.04 | 5.81 ± 0.01 |
| Q4 | 16.4 to <20.6 g/d | 236 | 26.0 ± 0.37 | 35.6 ± 0.50 | 0.56 ± 0.006 | 210 | 1.13 ± 0.05 | 5.81 ± 0.01 |
| Q5 | 20.6–41.4 g/d | 145 | 27.6 ± 0.56 | 38.0 ± 0.78 | 0.59 ± 0.010 | 128 | 1.24 ± 0.08 | 5.82 ± 0.02 |
| P-trend4 | 0.43 | 0.30 | 0.24 | 0.68 | 0.18 | |||
Values are mean ± SE unless otherwise indicated. CRP, C-reactive protein; HEI, Healthy Eating Index; Q, quintile.
Adjusted for age, HEI-2015, and weight-adjusted intakes of carbohydrates and MUFAs.
Adjusted for age, HEI-2015, weight-adjusted intakes of carbohydrates and MUFAs, and baseline BMI.
Means were derived from ANCOVA models with P values for linear trend derived from the model.
Table 2 also shows the associations between quintiles of saturated fat (from dairy and nondairy sources) and log-transformed concentrations of CRP and fibrinogen. After adjusting for BMI and other risk factors, dairy-derived saturated fat intakes were found to be inversely associated with both CRP and fibrinogen among males, but not females. For nondairy-derived saturated fats, males in higher quintiles of intake had slightly higher concentrations of fibrinogen than those in the lowest quintile of intake (P = 0.06).
Table 3 shows models evaluating the association between saturated fats from both dairy and nondairy sources and subsequent lipid concentrations; these models controlled for confounding by age, sex, diet quality (HEI-2015), weight-adjusted intakes of carbohydrates and MUFAs, and BMI. Here, males in higher quintiles of dairy-derived saturated fats had higher HDL-cholesterol concentrations than those in lower quintiles (P = 0.04), whereas TG concentrations declined with higher intakes of saturated fat from dairy sources (P < 0.01). Further, those males in the highest quintile of dairy-derived saturated fat had the lowest TG:HDL ratio whereas those in quintile 1 had the highest TG:HDL ratio (3.07 and 4.26, respectively). No such associations were observed with intakes of saturated fats from nondairy sources except for an inverse association with the TG:HDL ratio in females.
TABLE 3.
Sex-specific associations between quintiles of saturated fat from dairy and nondairy sources and mean lipid concentrations after 4 y of follow up1
| Type of fat | Range | n | HDL-C, mg/dL | LDL-C, mg/dL | Log TG,2 mg/dL | TG:HDL2 |
|---|---|---|---|---|---|---|
| Males | ||||||
| Dairy-derived saturated fat intake | ||||||
| Q1 | 0.0 to <4.5 g/d | 158 | 42.6 ± 1.02 | 126.1 ± 2.64 | 4.93 ± 0.05 | 4.26 ± 0.23 |
| Q2 | 4.5 to <7.1 g/d | 154 | 43.0 ± 0.98 | 130.5 ± 2.53 | 4.80 ± 0.05 | 3.60 ± 0.22 |
| Q3 | 7.1 to <9.9 g/d | 151 | 43.5 ± 0.97 | 126.2 ± 2.50 | 4.76 ± 0.04 | 3.30 ± 0.22 |
| Q4 | 9.9 to <13.7 g/d | 184 | 43.5 ± 0.88 | 127.7 ± 2.27 | 4.79 ± 0.04 | 3.68 ± 0.19 |
| Q5 | 13.7–40.4 g/d | 262 | 45.4 ± 0.80 | 129.3 ± 2.08 | 4.70 ± 0.04 | 3.07 ± 0.18 |
| P-trend3 | 0.04 | 0.63 | <0.01 | <0.01 | ||
| Nondairy saturated fat intake | ||||||
| Q1 | 0.0 to <10.5 g/d | 118 | 43.5 ± 1.45 | 121.0 ± 3.74 | 4.66 ± 0.07 | 3.02 ± 0.32 |
| Q2 | 10.5 to <13.4 g/d | 123 | 43.4 ± 1.17 | 129.7 ± 3.01 | 4.79 ± 0.05 | 3.61 ± 0.26 |
| Q3 | 13.4 to <16.4 g/d | 166 | 43.7 ± 0.96 | 129.8 ± 2.47 | 4.77 ± 0.04 | 3.62 ± 0.21 |
| Q4 | 16.4 to <20.6 g/d | 207 | 44.0 ± 0.83 | 129.2 ± 2.14 | 4.81 ± 0.04 | 3.48 ± 0.19 |
| Q5 | 20.6–41.4 g/d | 295 | 44.0 ± 0.98 | 128.5 ± 2.52 | 4.82 ± 0.05 | 3.69 ± 0.22 |
| P-trend3 | 0.69 | 0.32 | 0.18 | 0.35 | ||
| Females | ||||||
| Dairy-derived saturated fat intake | ||||||
| Q1 | 0.0 to <4.5 g/d | 207 | 59.0 ± 1.15 | 126.4 ± 2.47 | 4.69 ± 0.03 | 2.35 ± 0.12 |
| Q2 | 4.5 to <7.1 g/d | 267 | 58.6 ± 0.96 | 126.5 ± 2.08 | 4.71 ± 0.03 | 2.45 ± 0.10 |
| Q3 | 7.1 to <9.9 g/d | 248 | 61.0 ± 0.97 | 126.7 ± 2.10 | 4.70 ± 0.03 | 2.30 ± 0.10 |
| Q4 | 9.9 to <13.7 g/d | 251 | 61.2 ± 0.99 | 127.0 ± 2.13 | 4.64 ± 0.03 | 2.19 ± 0.11 |
| Q5 | 13.7–40.4 g/d | 184 | 58.7 ± 1.23 | 128.2 ± 2.65 | 4.70 ± 0.04 | 2.42 ± 0.13 |
| P-trend3 | 0.45 | 0.66 | 0.59 | 0.64 | ||
| Nondairy saturated fat intake | ||||||
| Q1 | 0.0 to <10.5 g/d | 271 | 58.8 ± 1.28 | 128.4 ± 2.75 | 4.69 ± 0.04 | 2.43 ± 0.14 |
| Q2 | 10.5 to <13.4 g/d | 282 | 58.4 ± 0.97 | 128.6 ± 2.09 | 4.76 ± 0.03 | 2.61 ± 0.10 |
| Q3 | 13.4 to <16.4 g/d | 246 | 59.2 ± 0.99 | 124.8 ± 2.12 | 4.68 ± 0.03 | 2.37 ± 0.10 |
| Q4 | 16.4 to <20.6 g/d | 219 | 61.8 ± 1.19 | 123.3 ± 2.55 | 4.64 ± 0.04 | 2.06 ± 0.13 |
| Q5 | 20.6–41.4 g/d | 139 | 62.2 ± 1.82 | 130.0 ± 3.92 | 4.64 ± 0.05 | 2.00 ± 0.19 |
| P-trend3 | 0.10 | 0.48 | 0.20 | 0.03 | ||
Values are mean ± SE, adjusted for age, BMI, Healthy Eating Index-2015, and weight-adjusted intakes of carbohydrates and MUFAs, unless otherwise indicated. HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; Q, quintile; TG, triglyceride.
Of the 2066 with other lipid values, 15 TG concentrations >500 mg/dL were excluded from the TG-related analyses.
Means were derived from ANCOVA models with P values for linear trend derived from the model.
Table 4 shows cross-sectional associations between saturated fats and lipid particle concentrations and sizes separately for males and females. Males with higher intakes of dairy-derived saturated fats had higher concentrations of HDL particles (P < 0.01) and fewer VLDL particles (P < 0.01). This was not the case in association with saturated fats from nondairy sources. In Table 4, it is also evident that HDL and LDL particle sizes were generally larger in females than in males. Among males, both HDL (P < 0.01) and LDL (P < 0.01) particle sizes were directly associated with quintiles of dairy-derived saturated fat, whereas no such association was observed for saturated fat from nondairy sources. Among females, there was a positive association of LDL particle size (P = 0.05) with dairy-derived saturated fat intakes, whereas saturated fats from nondairy sources showed a positive association with HDL particle size (P = 0.05).
TABLE 4.
Sex-specific cross-sectional associations between quintiles of saturated fat from dairy and nondairy sources and lipid particle sizes and concentrations1
| Lipid particle concentrations, mg/dL | Particle sizes, nm diameter | |||||||
|---|---|---|---|---|---|---|---|---|
| Type of fat | Range | n | HDL-P | LDL-P | Ln(VLDL-P)2 | HDL size | LDL size | VLDL size |
| Males | ||||||||
| Dairy-derived saturated fat intake | ||||||||
| Q1 | 0.0 to <4.5 g/d | 157 | 42.0 ± 0.90 | 134.7 ± 2.69 | 4.56 ± 0.06 | 8.84 ± 0.03 | 20.50 ± 0.05 | 49.30 ± 0.78 |
| Q2 | 4.5 to <7.1 g/d | 149 | 43.7 ± 0.87 | 136.3 ± 2.59 | 4.42 ± 0.05 | 8.88 ± 0.03 | 20.68 ± 0.05 | 47.67 ± 0.75 |
| Q3 | 7.1 to <9.9 g/d | 157 | 43.1 ± 0.83 | 133.1 ± 2.47 | 4.42 ± 0.05 | 8.91 ± 0.03 | 20.66 ± 0.05 | 47.43 ± 0.72 |
| Q4 | 9.9 to <13.7 g/d | 168 | 43.8 ± 0.81 | 133.0 ± 2.40 | 4.43 ± 0.05 | 8.88 ± 0.03 | 20.70 ± 0.05 | 48.86 ± 0.70 |
| Q5 | 13.7–40.4 g/d | 230 | 46.3 ± 0.76 | 131.7 ± 2.25 | 4.29 ± 0.05 | 8.98 ± 0.03 | 20.80 ± 0.04 | 46.97 ± 0.66 |
| P-trend3 | <0.01 | 0.28 | <0.01 | <0.01 | <0.01 | 0.16 | ||
| Nondairy saturated fat intake | ||||||||
| Q1 | 0.0 to <10.5 g/d | 116 | 44.4 ± 1.31 | 134.1 ± 3.89 | 4.38 ± 0.08 | 8.94 ± 0.05 | 20.75 ± 0.07 | 47.27 ± 1.14 |
| Q2 | 10.5 to <13.4 g/d | 120 | 45.0 ± 1.05 | 134.6 ± 3.12 | 4.44 ± 0.07 | 8.96 ± 0.04 | 20.71 ± 0.06 | 48.02 ± 0.91 |
| Q3 | 13.4 to <16.4 g/d | 160 | 43.7 ± 0.86 | 131.3 ± 2.54 | 4.41 ± 0.05 | 8.93 ± 0.03 | 20.73 ± 0.05 | 48.27 ± 0.74 |
| Q4 | 16.4 to <20.6 g/d | 190 | 43.6 ± 0.76 | 134.2 ± 2.26 | 4.45 ± 0.05 | 8.89 ± 0.03 | 20.66 ± 0.04 | 48.26 ± 0.66 |
| Q5 | 20.6–41.4 g/d | 275 | 43.8 ± 0.92 | 133.8 ± 2.71 | 4.39 ± 0.06 | 8.86 ± 0.03 | 20.63 ± 0.05 | 47.86 ± 0.79 |
| P-trend3 | 0.50 | 0.95 | 0.92 | 0.14 | 0.25 | 0.72 | ||
| Females | ||||||||
| Dairy-derived saturated fat intake | ||||||||
| Q1 | 0.0 to <4.5 g/d | 195 | 56.9 ± 1.03 | 127.7 ± 2.55 | 4.10 ± 0.05 | 9.42 ± 0.03 | 21.04 ± 0.04 | 44.40 ± 0.64 |
| Q2 | 4.5 to <7.1 g/d | 224 | 57.1 ± 0.91 | 127.3 ± 2.26 | 4.06 ± 0.04 | 9.41 ± 0.03 | 21.05 ± 0.03 | 43.94 ± 0.56 |
| Q3 | 7.1 to <9.9 g/d | 220 | 58.6 ± 0.89 | 129.9 ± 2.21 | 4.12 ± 0.04 | 9.41 ± 0.03 | 21.12 ± 0.03 | 43.94 ± 0.55 |
| Q4 | 9.9 to <13.7 g/d | 224 | 58.4 ± 0.91 | 126.8 ± 2.25 | 4.03 ± 0.04 | 9.44 ± 0.03 | 21.13 ± 0.03 | 44.36 ± 0.56 |
| Q5 | 13.7–40.4 g/d | 150 | 57.6 ± 1.18 | 129.8 ± 2.94 | 4.09 ± 0.06 | 9.36 ± 0.04 | 21.12 ± 0.04 | 44.61 ± 0.73 |
| P-trend3 | 0.37 | 0.77 | 0.72 | 0.69 | 0.05 | 0.74 | ||
| Nondairy saturated fat intake | ||||||||
| Q1 | 0.0 to <10.5 g/d | 250 | 57.3 ± 1.17 | 130.1 ± 2.89 | 4.13 ± 0.06 | 9.37 ± 0.04 | 21.06 ± 0.04 | 45.06 ± 0.72 |
| Q2 | 10.5 to <13.4 g/d | 253 | 57.0 ± 0.88 | 130.1 ± 2.17 | 4.15 ± 0.04 | 9.36 ± 0.03 | 21.04 ± 0.03 | 45.08 ± 0.54 |
| Q3 | 13.4 to <16.4 g/d | 207 | 58.1 ± 0.93 | 125.0 ± 2.30 | 4.01 ± 0.04 | 9.46 ± 0.03 | 21.15 ± 0.03 | 43.00 ± 0.57 |
| Q4 | 16.4 to <20.6 g/d | 192 | 58.4 ± 1.11 | 125.2 ± 2.75 | 4.07 ± 0.05 | 9.46 ± 0.03 | 21.13 ± 0.04 | 43.80 ± 0.68 |
| Q5 | 20.6–41.4 g/d | 111 | 58.9 ± 1.76 | 130.5 ± 4.37 | 3.97 ± 0.08 | 9.46 ± 0.05 | 21.10 ± 0.06 | 43.38 ± 1.09 |
| P-trend3 | 0.42 | 0.41 | 0.16 | 0.05 | 0.20 | 0.12 | ||
Values are mean ± SE, adjusted for age, BMI, Healthy Eating Index-2015, and weight-adjusted intakes of carbohydrates and MUFAs, unless otherwise indicated. HDL-P, high-density lipoprotein particle; LDL-P, low-density lipoprotein particle; Q, quintile; VLDL-P, very-low-density lipoprotein particle.
VLDL-P concentrations are ln-transformed.
Means were derived from ANCOVA models with P values for linear trend derived from the model.
Because all of the foregoing analyses were stratified by sex, statistical power for these stratified analyses may be lower. Supplemental Tables 3 and 4 show combined analyses for males and females. These results in general support beneficial associations between dairy-derived saturated fats and nearly all of these cardiometabolic outcomes (Supplemental Table 3). In contrast, most of the outcomes, except fibrinogen, were unassociated with nondairy-derived saturated fats (Supplemental Table 4).
Discussion
In this study, we found that higher mean intakes of saturated fat from dairy sources over 8 y were associated with lower levels of body fat in females. After accounting for differences in body fat and other risk factors among males, higher mean intakes of saturated fat from dairy sources were associated with a less atherogenic risk profile, including less inflammation, higher concentrations of HDL cholesterol, lower TG concentrations, and a lower TG:HDL ratio. Further, males in higher quintiles of saturated fat intake from dairy sources had higher HDL particle concentrations, lower VLDL concentrations, and larger HDL and LDL particle sizes. No such associations with lipids or inflammatory biomarkers were observed among females for intakes of dairy-derived saturated fats. Saturated fat from nondairy sources in females was inversely associated with the TG:HDL ratio.
A systematic review of the literature in 2013 concluded that high-fat dairy intake in most observational studies was not associated with BMI or risk of becoming overweight (7). Data from the Women's Health Study clinical trial also demonstrated that high-fat (but not low-fat) dairy consumption led to less weight gain and a lower risk of overweight and obesity in females (14). Further, a recent review of the evidence from other clinical trials concluded that saturated fat from dairy sources was either not associated or beneficially associated with weight change (15). These results are consistent with the current findings among females in this study (16, 17).
Some studies have found that selected dairy foods may attenuate inflammation and oxidative stress (18, 19), whereas others have concluded that dairy foods have no association with systemic inflammation (16). A large systematic review in 2017 used an inflammatory score to evaluate the evidence for anti-inflammatory and proinflammatory effects of dairy foods (20). Overall, this study confirmed a weak but important anti-inflammatory effect of dairy foods. However, they did find that dairy foods may have a proinflammatory effect among individuals with milk allergies. No sex-specific differences were discussed in that review. In the current study, we found anti-inflammatory effects of dairy-derived saturated fats although these beneficial effects were much stronger in males, suggesting that more studies are needed of sex-specific differences in the association between saturated fats and inflammation.
In this study, we found no association between saturated fat from any source and LDL-cholesterol concentrations. An earlier meta-analysis of randomized clinical trials also showed that higher consumption of both low-fat and high-fat dairy products had no statistically significant association with LDL-cholesterol concentrations (21). However, effects across individual studies were highly heterogeneous (and used heterogeneous study populations). A 2016 review also concluded that there was no evidence that milk adversely affected LDL-cholesterol concentrations (22), and a later summary of systematic reviews and meta-analyses of randomized trials also found no association between dairy consumption and LDL cholesterol (23). In a 2020 review, the authors concluded that reducing total intake of saturated fat led to small reductions in LDL cholesterol (24). However, this latter review focused on overall intake of saturated fat rather than saturated fat from different food sources as was done in the current study.
Several studies have found saturated fats in general to be positively associated with HDL cholesterol, although it is not clear whether this beneficial association was linked with dairy sources of saturated fat or total saturated fat (25, 26). There is other evidence, however, that dairy-derived saturated fats may be beneficial. For example, a controlled trial with high-fat cheese as an exposure variable found an increase in HDL cholesterol associated with the cheese-based intervention (27). An earlier meta-analysis of low-fat and high-fat dairy foods, however, found no effects on HDL cholesterol (21). Thus, the current study adds evidence that saturated fat from dairy sources may be beneficially associated with HDL cholesterol but more evidence is needed.
To our knowledge there are no studies of saturated fat from dairy compared with nondairy sources on TG concentrations. However, a randomized clinical trial of high-fat dairy in a Dietary Approaches to Stop Hypertension (DASH)-style diet found that the diet pattern led to significant reductions in TG concentrations. There are also studies showing variable effects of dairy intake on TGs (17, 28, 29). Dating back to the earliest release of the National Cholesterol Education Program (NCEP) Adult Treatment Panel III guidelines, scientists have emphasized the importance of considering combinations of lipids rather than individual lipid concentrations (30). The TG:HDL ratio that was examined in this study has been shown to be a strong predictor of both cardiometabolic health and all-cause mortality (31, 32). This is the first study, to our knowledge, to show that saturated fats from dairy foods were linked with a lower TG:HDL ratio. Another recent study, however, may conflict with our results. In that study of a so-called “Western Diet” (defined as high intakes of meats, processed foods, and high-fat dairy), investigators found that greater adherence to a “Western Diet” was associated with a higher risk of elevated TG concentrations and a higher TG:HDL ratio (33). However, in that study dairy fats cannot be separated from nondairy fats, making it impossible to determine whether all sources of fat in the “Western Diet” have comparable effects on lipids.
The association in the current study between saturated fats from dairy-derived sources and larger lipid particle sizes and lower concentrations of atherogenic particles has been previously observed (34). Interestingly, a study of low-fat dairy products found no association with lipid particle size, perhaps suggesting a preferential effect of higher-fat dairy on CVD risk via beneficial effects on particle size (29). Several studies have shown that fewer small, dense lipid particles are associated with a reduced risk of atherosclerosis (34) and ≥1 study concluded that a preponderance of small, dense LDL particles is particularly atherogenic even among those with normal LDL-cholesterol concentrations (35). Thus, it is possible that the beneficial association between dairy intake and LDL particle size (36) may explain some of the previously observed beneficial associations between total dairy intake and CVD risk (28, 37).
It is possible that the exposure variable in the current study, saturated fat from dairy sources, is a surrogate for higher dairy intakes in general, which may benefit CVD risk through other mechanisms. For example, it is also possible that the protein content of dairy may lower age-related weight gain and aid in maintenance of lean muscle mass, particularly among females who have lower baseline levels of lean mass. In addition, higher-fat dairy foods have been found to be inversely associated with glucose homeostasis in some studies (7), and a randomized clinical trial demonstrated that whey protein supplementation led to a lower postprandial chylomicron response among individuals with excess visceral fat (38). Further, the mineral content of dairy (calcium, potassium, magnesium, phosphorus) may have additional beneficial effects as well (39).
Saturated fat from dairy sources has a diverse fatty acid profile consisting of short-, medium-, long-, odd-, and branched-chain fatty acids, whereas meats contribute predominantly medium- and long-chain fatty acids to the American diet (15). It is very possible that the different chain lengths and different food matrices contributing to saturated fat intakes will have different effects on health outcomes. For example, in 1 study, SCFAs and medium-chain fatty acids were not associated with CVD risk (40). In another, higher intakes of medium-chain fatty acids were associated with lower total mortality in males and both medium- and odd-chain fatty acids were inversely associated with mortality in both males and females (41). Thus, it seems essential to consider the specific food sources of saturated fats to better understand the relevant health effects of various fatty acids.
This study has a number of strengths and limitations. First of all, the detailed dietary records, representing nearly 16,000 d of intake, are an important strength because dietary records are often considered a gold-standard method for dietary assessment, especially in larger epidemiologic studies (42). In addition, the careful linkage of all foods from the diet records with USDA MyPyramid serving equivalents is a strength. This allowed us to separate the saturated fat intake associated with dairy foods from that derived from other food sources. This provided us with a unique data resource with more accurate and detailed information on saturated fat from dairy and nondairy sources. Another important strength of the study is the carefully collected data on potential confounders of interest in the Framingham cohorts, thereby enhancing the precision and validity of the results. This study is of course not without limitations, starting with the fact that 35% of the participants were missing dietary data or were excluded on the basis of extreme intake values. Another 9% of participants died before the collection of dietary information and 8% were excluded for other reasons. Thus, the remaining participants in the analysis data set may differ from the underlying study cohort. All long-term studies of diet rely on self-reported data and this is no exception. Thus, there is always the possibility of biased reporting of intake. In this study, the dietary records were completed independently by the subjects although detailed debriefing was carried out as well with every subject. Nonetheless, random misclassification of dietary intake is also possible. The intercorrelations of dietary factors in observational studies make it difficult to separate out individual nutrient effects. Although we carried out extensive examination of potential confounding by other dietary factors, we cannot completely rule out confounding by diet and other factors. Lastly, the long-standing Framingham studies include almost exclusively Caucasian participants of European descent. These results may well vary in different racial/ethnic groups with very different dietary patterns and other risk factors.
In conclusion, this longitudinal study provides important new evidence that saturated fat from dairy sources was beneficially associated with markers of CVD risk, including anthropometric measures of body fat in females; biomarkers of inflammation, HDL cholesterol, TG, and the HDL:TG ratio in males; and lipid particles in both males and females in the Framingham Offspring cohort.
Supplementary Material
Acknowledgements
The authors’ responsibilities were as follows—LLM: designed the research; MRS: provided essential materials and provided oversight for data analysis; MY: analyzed the data; MY and LLM: wrote the paper and have primary responsibility for the final content; and all authors: conducted the research, interpreted the results, and read and approved the final manuscript. The authors report no conflicts of interest.
Notes
These data were originally collected with funding from the National Heart, Lung, and Blood Institute (Framingham Study contract N01-HC-25195 and HHSN268201500001I). Analyses were supported through National Dairy Council grant 2854 (to LLM). The funders had no role in the study design, data collection and analysis, interpretation of data, or writing of the report. There were no restrictions on submission of the report for publication and the funders were not provided with a copy of the manuscript before submission.
Supplemental Tables 1–4 are available from the “Supplementary data” link in the online posting of the article and from the same link in the online table of contents at https://academic.oup.com/ajcn/.
Abbreviations used: CRP, C-reactive protein; CVD, cardiovascular disease; DOS, disk operating system; HEI, Healthy Eating Index; NDSR, Nutrition Data System for Research; TG, triglyceride.
Contributor Information
Mengjie Yuan, Section of Preventive Medicine and Epidemiology, Department of Medicine, Boston University School of Medicine, Boston, MA, USA.
Martha R Singer, Section of Preventive Medicine and Epidemiology, Department of Medicine, Boston University School of Medicine, Boston, MA, USA.
Richard T Pickering, Section of Infectious Diseases, Department of Medicine, Boston University School of Medicine, Boston, MA, USA.
Lynn L Moore, Section of Preventive Medicine and Epidemiology, Department of Medicine, Boston University School of Medicine, Boston, MA, USA.
Data availability
Data described in the article, code book, and analytic code may be made available upon request pending application to and approval by the Framingham Executive Committee and/or the Framingham Research Committee, and payment of any necessary data fees.
References
- 1. Page IH, Allen EV, Chamberlain FL, Keys A, Stamler J, Stare FJ. Dietary fat and its relation to heart attacks and strokes. Report by the Central Committee for Medical and Community Program of the American Heart Association. JAMA. 1961;175(5):389–91. [PubMed] [Google Scholar]
- 2. DeSouza RJ, Mente A, Maroleanu A, Cozma AI, Ha V, Kishibe Tet al. Intake of saturated and trans unsaturated fatty acids and risk of all cause mortality, cardiovascular disease, and type 2 diabetes: systematic review and meta-analysis of observational studies. BMJ. 2015;351:h3978. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Houston M. The relationship of saturated fats and coronary heart disease: fa(c)t or fiction? A commentary. Ther Adv Cardiovasc Dis. 2018;12(2):33–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Ruiz-Núñez B, Dijck-Brouwer J, Muskiet FA. The relation of saturated fatty acids with low-grade inflammation and cardiovascular disease. J Nutr Biochem. 2016;36:1–20. [DOI] [PubMed] [Google Scholar]
- 5. USDA and US Department of Health and Human Services (DHHS) . Dietary Guidelines for Americans 2020–2025[Internet]. 9th ed. Washington (DC): USDA and US DHHS; 2020. [ cited July 12, 2022]. Available from: https://www.dietaryguidelines.gov. [Google Scholar]
- 6. Astrup A, Magkos F, Bier DM, Brenna JT, de Oliveira Otto MC, Hill JOet al. Saturated fats and health: a reassessment and proposal for food-based recommendations: JACC State-of-the-Art review. J Am Coll Cardiol. 2020;76(7):844–57. [DOI] [PubMed] [Google Scholar]
- 7. Kratz M, Baars T, Guyenet S. The relationship between high-fat dairy consumption and obesity, cardiovascular, and metabolic disease. Eur J Nutr. 2013;52(1):1–24. [DOI] [PubMed] [Google Scholar]
- 8. Mensink RP, Zock PL, Katan MB. Effects of dietary fatty acids and carbohydrates on the ratio of serum total to HDL cholesterol and on serum lipids and apolipoproteins: a meta-analysis of 60 controlled trials. Am J Clin Nutr. 2003;77(5):1146–55. [DOI] [PubMed] [Google Scholar]
- 9. Schakel SF, Sievert YA, Buzzard IM. Sources of data for developing and maintaining a nutrient database. J Am Diet Assoc. 1988;88(10):1268–71. [PubMed] [Google Scholar]
- 10. Lukaski H, Bolonchuk W, Hall C, Siders W. Validation of tetrapolar bioelectrical impedance method to assess human body composition. J Appl Physiol. 1986;60(4):1327–32. [DOI] [PubMed] [Google Scholar]
- 11. Friedewald WT, Levy RI, Fredrickson DS. Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge. Clin Chem. 1972;18(6):499–502. [PubMed] [Google Scholar]
- 12. Cromwell WC, Otvos JD, Keyes MJ, Pencina MJ, Sullivan L, Vasan RSet al. LDL particle number and risk of future cardiovascular disease in the Framingham Offspring Study—implications for LDL management. J Clin Lipidol. 2007;1(6):583–92. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Kannel WB, Sorlie P. Some health benefits of physical activity: the Framingham Study. Arch Intern Med. 1979;139(8):857–61. [PubMed] [Google Scholar]
- 14. Rautiainen S, Wang L, Lee I, Manson JE, Buring JE, Sesso HD. Dairy consumption in association with weight change and risk of becoming overweight or obese in middle-aged and older women: a prospective cohort study. Am J Clin Nutr. 2016;103(4):979–88. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Unger AL, Torres-Gonzalez M, Kraft J. Dairy fat consumption and risk of metabolic syndrome: an examination of the saturated fatty acids in dairy. Nutrients. 2019;11(9):2200. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Labonté M-È, Cyr A, Abdullah MM, Lépine M-C, Vohl M-C, Jones Pet al. Dairy product consumption has no impact on biomarkers of inflammation among men and women with low-grade systemic inflammation. J Nutr. 2014;144(11):1760–7. [DOI] [PubMed] [Google Scholar]
- 17. Wennersberg MH, Smedman A, Turpeinen AM, Retterstøl K, Tengblad S, Lipre Eet al. Dairy products and metabolic effects in overweight men and women: results from a 6-mo intervention study. Am J Clin Nutr. 2009;90(4):960–8. [DOI] [PubMed] [Google Scholar]
- 18. Yuan M, Singer MR, Moore LL. Yogurt consumption is associated with lower levels of chronic inflammation in the Framingham Offspring Study. Nutrients. 2021;13(2):506. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Stancliffe RA, Thorpe T, Zemel MB. Dairy attenuates oxidative and inflammatory stress in metabolic syndrome. Am J Clin Nutr. 2011;94(2):422–30. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Bordoni A, Danesi F, Dardevet D, Dupont D, Fernandez AS, Gille Det al. Dairy products and inflammation: a review of the clinical evidence. Crit Rev Food Sci Nutr. 2017;57(12):2497–525. [DOI] [PubMed] [Google Scholar]
- 21. Benatar JR, Sidhu K, Stewart RAH. Effects of high and low fat dairy food on cardio-metabolic risk factors: a meta-analysis of randomized studies. PLoS One. 2013;8(10):e76480. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Drouin-Chartier J-P, Côté JA, Labonté M-È, Brassard D, Tessier-Grenier M, Desroches Set al. Comprehensive review of the impact of dairy foods and dairy fat on cardiometabolic risk. Adv Nutr. 2016;7(6):1041–51. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Fontecha J, Visitación Calvo M, Juarez M, Gil A, Martínez-Vizcaino V. Milk and dairy product consumption and cardiovascular diseases: an overview of systematic reviews and meta-analyses. Adv Nutr. 2019;10(suppl_2):S164–89. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Hooper L, Martin N, Jimoh OF, Kirk C, Foster E, Abdelhamid AS. Reduction in saturated fat intake for cardiovascular disease. Cochrane Database Syst Rev. 2020;8(8):CD011737. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Abdullah M, Cyr A, Lépine M-C, Labonté M-È, Couture P, Jones Pet al. Recommended dairy product intake modulates circulating fatty acid profile in healthy adults: a multi-centre cross-over study. Br J Nutr. 2015;113(3):435–44. [DOI] [PubMed] [Google Scholar]
- 26. Chiu S, Bergeron N, Williams PT, Bray GA, Sutherland B, Krauss RM. Comparison of the DASH (Dietary Approaches to Stop Hypertension) diet and a higher-fat DASH diet on blood pressure and lipids and lipoproteins: a randomized controlled trial. Am J Clin Nutr. 2016;103(2):341–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Thorning TK, Raziani F, Bendsen NT, Astrup A, Tholstrup T, Raben A. Diets with high-fat cheese, high-fat meat, or carbohydrate on cardiovascular risk markers in overweight postmenopausal women: a randomized crossover trial. Am J Clin Nutr. 2015;102(3):573–81. [DOI] [PubMed] [Google Scholar]
- 28. Crichton GE, Alkerwi A. Dairy food intake is positively associated with cardiovascular health: findings from Observation of Cardiovascular Risk Factors in Luxembourg study. Nutr Res. 2014;34(12):1036–44. [DOI] [PubMed] [Google Scholar]
- 29. van Meijl LEC, Mensink RP. Effects of low-fat dairy consumption on markers of low-grade systemic inflammation and endothelial function in overweight and obese subjects: an intervention study. Br J Nutr. 2010;104(10):1523–7. [DOI] [PubMed] [Google Scholar]
- 30. Grundy SM, Becker D, Clark LT, Cooper RS, Denke MA, Howard WJet al. Third report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III) final report. Circulation. 2002;106(25):3143–421. [PubMed] [Google Scholar]
- 31. Bittner V, Johnson BD, Zineh I, Rogers WJ, Vido D, Marroquin OCet al. The TG/HDL cholesterol ratio predicts all cause mortality in women with suspected myocardial ischemia: a report from the Women's Ischemia Syndrome Evaluation (WISE). Am Heart J. 2009;157(3):548–55. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Salazar MR, Carbajal HA, Espeche WG, Leiva Sisnieguez CE, Balbín E, Dulbecco CAet al. Relation among the plasma triglyceride/high-density lipoprotein cholesterol concentration ratio, insulin resistance, and associated cardio-metabolic risk factors in men and women. Am J Cardiol. 2012;109(12):1749–53. [DOI] [PubMed] [Google Scholar]
- 33. Ushula TW, Mamun A, Darssan D, Wang WYS, Williams GM, Whiting SJet al. Dietary patterns and the risk of abnormal blood lipids among young adults: a prospective cohort study. Nutr Metab Cardiovasc Dis. 2022;32(5):1165–74. [DOI] [PubMed] [Google Scholar]
- 34. Sjogren P, Rosell M, Skoglund-Andersson C, Zdravkovic S, Vessby B, de Faire Uet al. Milk-derived fatty acids are associated with a more favorable LDL particle size distribution in healthy men. J Nutr. 2004;134(7):1729–35. [DOI] [PubMed] [Google Scholar]
- 35. Hulthe J, Bokemark L, Wikstrand J, Fagerberg B. The metabolic syndrome, LDL particle size, and atherosclerosis: the Atherosclerosis and Insulin Resistance (AIR) study. Arterioscler Thromb Vasc Biol. 2000;20(9):2140–7. [DOI] [PubMed] [Google Scholar]
- 36. Superko HR, Gadesam RR. Is it LDL particle size or number that correlates with risk for cardiovascular disease?. Curr Atheroscler Rep. 2008;10(5):377–85. [DOI] [PubMed] [Google Scholar]
- 37. Qin L-Q, Xu J-Y, Han S-F, Zhang Z-L, Zhao Y-Y, Szeto IM. Dairy consumption and risk of cardiovascular disease: an updated meta-analysis of prospective cohort studies. Asia Pac J Clin Nutr. 2015;24(1):90–100. [DOI] [PubMed] [Google Scholar]
- 38. Bohl M, Bjørnshave A, Rasmussen KV, Grethe Schioldan A, Amer B, Larsen MKet al. Dairy proteins, dairy lipids, and postprandial lipemia in persons with abdominal obesity (DairyHealth): a 12-wk, randomized, parallel-controlled, double-blinded, diet intervention study. Am J Clin Nutr. 2015;101(4):870–8. [DOI] [PubMed] [Google Scholar]
- 39. Weaver CM. How sound is the science behind the dietary recommendations for dairy?. Am J Clin Nutr. 2014;99(5):1217S–22S. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40. Hu FB, Stampfer MJ, Manson JE, Ascherio A, Colditz GA, Speizer FEet al. Dietary saturated fats and their food sources in relation to the risk of coronary heart disease in women. Am J Clin Nutr. 1999;70(6):1001–8. [DOI] [PubMed] [Google Scholar]
- 41. Zhuang P, Cheng L, Wang J, Zhang Y, Jiao J. Saturated fatty acid intake is associated with total mortality in a nationwide cohort study. J Nutr. 2019;149(1):68–77. [DOI] [PubMed] [Google Scholar]
- 42. Park Y, Dodd KW, Kipnis V, Thompson FE, Potischman N, Schoeller DAet al. Comparison of self-reported dietary intakes from the Automated Self-Administered 24-h recall, 4-d food records, and food-frequency questionnaires against recovery biomarkers. Am J Clin Nutr. 2018;107(1):80–93. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Data described in the article, code book, and analytic code may be made available upon request pending application to and approval by the Framingham Executive Committee and/or the Framingham Research Committee, and payment of any necessary data fees.