The Portfolio Diet Score and Risk of Cardiovascular Disease: Findings from Three Prospective Cohort Studies

Abstract

Background:

The plant-based portfolio dietary pattern includes recognized cholesterol-lowering foods (plant protein, nuts, viscous fiber, phytosterols, plant monounsaturated fats [MUFAs]) shown to improve several cardiovascular disease (CVD) risk factors in randomized controlled trials. However, there is limited evidence on the role of long-term adherence to the diet and CVD risk. The primary objective was to examine the relation between the portfolio diet score (PDS) and the risk of total CVD, coronary heart disease (CHD), and stroke.

Methods:

We prospectively followed 73,924 women in the Nurses’ Health Study (NHS) (1984–2016), 92,346 women in the NHSII (1991–2017) and 43,970 men from the Health Professionals Follow-up Study (HPFS) (1986–2016) without CVD and cancer at baseline. Diet was assessed using validated food frequency questionnaires at baseline and every four years using a PDS that positively ranks plant protein (legumes), nuts and seeds, viscous fiber sources, phytosterols (mg/day) and plant MUFA sources, and negatively ranks foods high in saturated fat and cholesterol.

Results:

During up to 30 years of follow-up, 16,917 incident CVD cases, including 10,666 CHD cases and 6,473 strokes, were documented. After multivariable adjustment for lifestyle factors and a modified Alternate Healthy Eating Index (excluding overlapping components), comparing the highest to the lowest quintile, participants with a higher PDS had a lower risk of total CVD (pooled HR: 0.86; 95% CI: 0.81–0.92, P trend<0.001), CHD (pooled HR: 0.86; 95% CI: 0.80–0.93, P trend=0.0001) and stroke (pooled HR: 0.86; 95% CI: 0.78–0.95, P trend=0.0003). In addition, a 25-percentile higher PDS was associated with a lower risk of total CVD (pooled HR: 0.92; 95% CI: 0.89–0.95), CHD (pooled HR: 0.92; 95% CI: 0.88–0.95) and stroke (pooled HR: 0.92; 95% CI: 0.87–0.96). Results remained consistent across sensitivity and most subgroup analyses, and there was no evidence of departure from linearity for CVD, CHD or stroke. In a subset of participants, a higher PDS was associated with a more favorable blood lipid and inflammatory profile.

Conclusions:

The PDS was associated with a lower risk of CVD, including CHD and stroke, and a more favorable blood lipid and inflammatory profile, in three large prospective cohorts.

INTRODUCTION

Cardiovascular disease (CVD) continues to be a leading cause of death in the United States and globally¹. Multiple lines of evidence, including randomized controlled trials (RCTs) and Mendelian randomization studies, have consistently established the causal role of atherogenic blood lipids, low-density lipoprotein cholesterol (LDL-C), non-high-density lipoprotein cholesterol (non-HDL-C), and apolipoprotein B (ApoB), in the pathogenesis of atherosclerotic CVD^2–4. In the prevention and management of CVD, diet is a crucial modifiable risk factor⁵. A dietary approach that has demonstrated efficacy in reducing these atherogenic blood lipids is the portfolio dietary pattern, as evidenced by significant reductions in LDL-C, non-HDL-C, and ApoB in RCTs⁶. This plant-based dietary pattern combines well-established cholesterol-lowering foods and nutrients: the diet is low in foods high in saturated fat and cholesterol and encourages plant protein (legumes, particularly soy), viscous fiber sources (oats, barley, psyllium, okra, eggplant, berries, apples and citrus fruit), nuts and seeds, phytosterols (through fortified foods or supplements), and plant monounsaturated fat (MUFA) sources (e.g., from healthy unsaturated plant oils and avocado)^{7, 8}. Findings from an early metabolically controlled RCT showed that the LDL-C lowering effect of the portfolio diet was similar to 20mg lovastatin (−29% versus −31%)⁹. A recent systematic review and meta-analysis of all metabolically controlled and ad libitum trials confirmed these effects and showed that the diet significantly lowered LDL-C by 17%, non-HDL-C by 14%, and ApoB by 15%, as well as high-sensitivity C-reactive protein (hsCRP) by 32%, compared to a low saturated fat diet, independent of significant weight loss⁶.

Although RCTs have reported significant reductions in intermediate risk factors for CVD over the short-term (four weeks to six months), there is limited evidence on the role of long-term adherence to the portfolio dietary pattern and CVD risk. One observational cohort study found that the portfolio diet score (PDS) was associated with a lower risk of CVD, particularly coronary heart disease (CHD), in postmenopausal women¹⁰, however, these findings need to be confirmed in other populations. In addition, there is limited evidence quantifying the CVD risk-lowering potential of the portfolio dietary pattern that combines these well-established cholesterol-lowering foods.

Therefore, this study examined the PDS and its association with incident CVD, CHD and stroke in three large U.S. prospective cohorts of men and women. This study further examined associations with plasma levels of lipid and inflammatory biomarkers in a subpopulation of the cohorts.

METHODS

Study Population:

The three prospective cohort studies include the Nurses’ Health Study (NHS), NHSII and Health Professionals Follow-up Study (HPFS)^{11, 12}. The NHS began in 1976, with 121,701 female nurses aged 30 to 55 years recruited. The NHSII started in 1989 and includes 116,429 female nurses aged 25 to 42 years. The HPFS started in 1986, recruiting 51,529 male health professionals aged 40 to 75 years. Participants in each cohort were followed up biennially on lifestyle, medical history, and other health-related factors, with a response rate of ~90%. Participants who had cancer, CVD, those who did not complete FFQs or met the FFQ exclusion criteria (missing dietary data or a reported energy intake of <600 or >3500 kcal/day for women and <800 or >4200 kcal/day for men) at baseline in each cohort (1984 for NHS, 1991 for NHSII and 1986 for HPFS) were excluded. The final analysis included 73,924 women from NHS, 92,346 women from NHSII and 43,970 men from HPFS. The study protocol was approved by the Institutional Review Boards of Brigham and Women’s Hospital and the Harvard T.H. Chan School of Public Health and informed consent was implied by the return of the cohort questionnaires. The data that support the findings of this study are available from the corresponding author upon reasonable request.

Assessment of the Portfolio Dietary Pattern:

A validated semi-quantitative FFQ was administered every 4 years to determine dietary intake, starting in 1984 and 1986 in NHS, 1991 in NHSII and 1986 in HPFS^13–15. The portfolio dietary pattern was assessed using a PDS that was previously developed and validated using a modified FFQ used in the current cohorts against diet records and LDL-C¹⁶. The correlation between the PDS from the FFQ and diet records was 0.69, highlighting reasonable validity. Foods on the FFQs were categorized into the six components of the portfolio diet (plant protein [legumes, beans, tofu, peas, soymilk], nuts and seeds [nuts, peanut butter, flaxseed], viscous fiber sources [Bran buds, oats, oat bran, apples, applesauce, citrus fruit, berries, eggplant], phytosterols [estimated from all foods], plant MUFA sources [avocado, olive oil, canola oil, soy oil], and high saturated fat and cholesterol sources [whole fat dairy, eggs with yolk, red and processed meats, organ meat, butter, etc.]). Intake was assessed as servings/day reported from the FFQ of targeted foods except phytosterols, which was based on all FFQ items to derive total intake (mg/day). For the six components, each was scored from 1 (least adherent) to 5 (most adherent) according to participant’s quintile of intake resulting in a score range between 6 and 30. A higher score indicates higher consumption of foods recommended in the diet. Average intakes per quintile are shown in Table S1. The PDS was determined using the dietary data in each of the 4-year FFQ cycles and the primary exposure included computing cumulatively averaged scores at each cycle to best represent long-term diet and to help reduce measurement error as the primary exposure. As participants may have altered their diet after diagnosis of a major illness, the dietary variables were not updated when participants reported a diagnosis of coronary revascularization, diabetes, angina, or cancer.

Ascertainment of CVD Outcomes:

CVD was defined as a composite of incident nonfatal myocardial infarction (MI), fatal CHD and fatal and nonfatal stroke. Nonfatal MI was confirmed by physicians according to the World Health Organization criteria¹⁷ including diagnostic electrocardiographic changes or elevated cardiac enzymes and nonfatal stroke was confirmed according to the National Survey of Stroke criteria¹⁸. Death was identified from next of kin, postal authorities or the National Death Index¹⁹. Cause of death was defined according to the International Classification of Diseases-8^th Revision. Fatal CHD and stroke were confirmed by autopsy records or death certificate/other evidence.

Plasma Biomarker Assessment:

Blood samples were collected in subpopulations in the NHS (n = 32,862) between 1989 and 1990, NHSII (n = 29, 611) between 1996 to 1999, and HPFS (n= 18,019) between 1993 to 1995²⁰. Plasma levels of total cholesterol (Total-C), LDL-C, triglycerides, high-density lipoprotein cholesterol (HDL-C), interleukin-6 (IL-6), soluble intercellular adhesion molecule-1 (sICAM-1), tumor necrosis factor-α receptor 1 (TNFα-RI), tumor necrosis factor-α receptor 2 (TNFα-R2), hsCRP, adiponectin and leptin were measured in several nested case-control studies of chronic disease within each cohort. Non-HDL-C was calculated by subtracting HDL-C from Total-C. Data were combined from these sub studies and corrected for batch effects using an average batch correction method²¹. Outliers, duplicates, individuals with missing diet data and cancer or CVD at blood draw were excluded, with a final sample size of 4,375 to 21,834 for individual biomarkers.

Covariate Assessment:

Updated information on the participants’ medical history, family history, lifestyle, body mass index (BMI), reproductive factors and medication use were determined through biennial questionnaires. Fasting status and corticosteroid use at blood draw were obtained from questionnaires completed by participants near the time of blood sample collection.

Statistical Analyses:

Person-years of follow-up were calculated from the return date of the first FFQ until date of CVD diagnosis, death or end of follow-up (June 2016 in NHS, June 2017 in NHSII and January 2016 in HPFS), whichever came first. Cox regressions with time-varying covariates were used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) of CVD outcomes (total CVD, CHD, and stroke), comparing high to low quintiles of the PDS. The PDS was also analyzed as a continuous variable, per a 25-percentile increment (i.e. per 6 points). To quantify a linear trend, the median value of the PDS within each quintile was assigned and the exposure was modeled as a continuous variable. Restricted cubic spline plots with four knots were used to explore the shape of the association between the PDS and total CVD, CHD and stroke. P-values for non-linearity and a linear relation were determined from the likelihood ratio tests. Further analyses were conducted to assess stroke subtypes, including ischemic and hemorrhagic stroke. All analyses were stratified by age and follow-up intervals, and multivariable models were further adjusted for race, smoking, menopausal status and postmenopausal hormone use (in women), oral contraceptive use (in women), multivitamin use, regular aspirin use, physical activity, family history of MI, family history of diabetes, marital status, BMI, total energy intake, alcohol intake, a modified Alternate Healthy Eating Index (AHEI [with components removed that have significant overlap with the PDS: fruit, nuts and legumes, polyunsaturated fats, and red and processed meat; alcohol was also removed]), and baseline hypercholesterolemia (includes cholesterol-lowering medications), hypertension (includes blood pressure-lowering medications) and diabetes status. Analyses were performed separately in each cohort and pooled HRs were obtained by combining data from the three cohorts.

Several sensitivity analyses were conducted. First, to account for potential confounding by socioeconomic status (SES), the models were adjusted for census-tract median family income, median home value, and percentage with a college degree. Second, the models were adjusted for updated diagnoses of intermediate diseases (hypercholesterolemia or medication, hypertension or medication and diabetes). Third, instead of using the cumulative average diet, the most recent measure of diet with CVD risk was used (simple update). Fourth, the cumulative average diet was continuously updated until the end of follow-up rather than stop updating when an intermediate disease occurred (hypercholesterolemia or medication, hypertension or medication and diabetes). A priori stratified analyses included age, BMI, sex, family history of MI, physical activity level, smoking, hypertension and hypercholesterolemia for total CVD, CHD and stroke based on a 25% increment increase in the PDS. Post hoc stratified analysis included by race/ethnicity. Lastly, the potential role of BMI in mediating the PDS CVD associations was examined.

Linear regressions were used to analyze associations of the PDS with levels of the lipid and inflammatory biomarkers. The average PDS from the 2 FFQs administered several years before the blood collection were used to help capture long-term diet and minimize reverse causality concerns (1984 & 1986 in NHS, 1991 & 1995 in NHSII and 1986 & 1990 in HPFS). The multivariable models were adjusted for the above covariates plus fasting status, corticosteroid use, study cohort and case-control status from the sub studies. Additional analyses included assessing the individual components of the PDS (plant protein, viscous fiber sources, nuts, phytosterols, MUFAs and saturated fat and cholesterol sources) individually. Spearman correlation coefficients were used to assess how correlated the PDS was with other dietary pattern indices (plant-based diet index [PDI], healthy plant-based diet index [hPDI], unhealthy plant-based diet index [uPDI], AHEI, Alternate Mediterranean [aMED] and Dietary Approaches to Stop Hypertension [DASH]). The proportional hazards assumption was evaluated with a likelihood ratio test comparing the model with and without an interaction term between age and the PDS, and the tests did not indicate a violation in any cohort. Analyses were performed with the SAS statistical package 9.4 (SAS Institute, Cary, North Carolina). Statistical tests were 2-sided, and p values of <0.05 considered statistically significant.

RESULTS

Participant Characteristics

During up to 30 years of follow-up, 16,917 incident CVD cases, including 10,666 CHD cases and 6,473 strokes, were documented. Baseline characteristics of the participants by cohort according to quintile of the PDS are presented in Table 1. Participants with a higher PDS reported being more physically active, had a higher energy intake, were more likely to take multivitamins and were less likely to smoke. Mean intake of the PDS components by cohort are shown in Table S1. The Spearman correlations between the PDS and other dietary patterns are shown in Table S2. The PDS was most closely correlated with the aMED (r = 0.78 to 0.81 across cohorts), followed by DASH (r = 0.65 to 0.77) and PDI (r = 0.62 to 0.67).

Table 1:

Baseline Characteristics of Participants According to Quintiles of the Portfolio Diet Score (PDS)

Variables	Q1	Q2	Q3	Q4	Q5
NHS (1984)
No. of participants	16,444	11,136	18,633	11,099	16,612
Dietary score, mean (SD)	12.2 (1.7)	15.5 (0.5)	18 (0.8)	20.5 (0.5)	23.7 (1.7)
Age, mean (SD), y	49.4 (7.1)	49.9 (7.1)	50.1 (7.1)	50.4 (7.1)	51.1 (7.1)
Non-Hispanic white race/ethnicity	97.8	98.1	97.8	97.9	97.5
Body mass index, mean (SD), kg/m2	25.0 (4.9)	24.9 (4.7)	25.0 (4.7)	25.0 (4.6)	25.0 (4.6)
Physical activity, mean (SD), MET-h/wk	11.5 (18.4)	12.6 (19.6)	13.8 (20.8)	15.4 (21.8)	17.2 (23.5)
Smoker, current	33.0	27.7	23.6	20.1	16.4
Alcohol, mean (SD), g/day	7.4 (12.1)	7.1 (11.7)	6.9 (11.4)	6.7 (10.4)	6.5 (10.2)
Diabetes	2.2	2.3	2.3	2.4	2.5
Family history of MI	24.9	25.0	24.8	25.8	26.0
Total energy intake, mean (SD), kcal/d	1419 (427)	1560 (453)	1731 (471)	1904 (497)	2103 (507)
Multivitamin use	33.1	35.4	36.3	38.3	41.5
Aspirin use	69.6	70.0	71.5	72.3	72.0
NHSII (1991)
No. of participants	20,931	13,587	22,034	19,306	16,488
Dietary score, mean (SD)	12.2 (1.7)	15.5 (0.5)	18 (0.8)	20.9 (0.8)	24.6 (1.6)
Age, mean (SD), y	35.8 (4.8)	35.9 (4.7)	36.1 (4.6)	36.2 (4.6)	36.6 (4.5)
Non-Hispanic white race/ethnicity	95.8	96.7	96.7	96.7	96.5
Body mass index, mean (SD), kg/m2	24.8 (5.6)	24.7 (5.3)	24.6 (5.3)	24.6 (5.2)	24.4 (5.1)
Physical activity, mean (SD), MET-h/wk	16.4 (23.9)	18.4 (24.2)	20.3 (25.7)	22.4 (27.7)	27.3 (33.1)
Smoker, current	17.0	13.5	11.8	9.7	8.2
Alcohol, mean (SD), g/day	2.9 (6.3)	3.1 (6.3)	3 (5.8)	3.2 (6)	3.3 (6)
Diabetes	0.8	1.1	1.0	1.0	1.0
Family history of MI	20.8	20.6	20.0	20.0	19.7
Total energy intake, mean (SD), kcal/d	1433 (437)	1594 (458)	1769 (478)	1992 (507)	2194 (508)
Multivitamin use	37.2	41.2	43.8	47.1	50.9
Aspirin use	11.7	11.0	11.0	11.1	11.1
HPFS (1986)
No. of participants	7,445	9,361	10,714	6,483	9,967
Dietary score, mean (SD)	11.5 (1.5)	15.1 (0.8)	18 (0.8)	20.5 (0.5)	23.8 (1.7)
Age, mean (SD), y	52.5 (9.6)	52.6 (9.4)	53.1 (9.5)	53.5 (9.6)	54.2 (9.6)
Non-Hispanic white race/ethnicity	94.6	94.5	95.2	95.2	95.0
Body mass index, mean (SD), kg/m2	25.7 (3.4)	25.6 (3.4)	25.5 (3.2)	25.4 (3.2)	25.2 (3.4)
Physical activity, mean (SD), MET-h/wk	15.6 (21.5)	17.9 (22.5)	19.7 (24)	21.9 (25.8)	25.7 (28.7)
Smoker, current	14.3	11.5	8.6	7.1	5.4
Alcohol, mean (SD), g/day	11.7 (16.3)	11.9 (16.1)	11.5 (15.1)	11.3 (15.5)	10.8 (14.5)
Diabetes	2.3	2.5	2.3	3.1	2.6
Family history of MI	14.5	14.8	15.1	14.8	14.8
Total energy intake, mean (SD), kcal/d	1604 (469)	1770 (524)	1976 (560)	2178 (586)	2398 (618)
Multivitamin use	37.7	39.8	41.3	42.6	46.9
Aspirin use	25.5	25.5	27.5	27.4	27.5

Values are mean ±SD or % and are standardized to the age distribution of the study population. Q=quintile; MI=myocardial infarction, MET=metabolic equivalent task

Portfolio dietary pattern and risk of CVD outcomes:

After multivariable adjustment for lifestyle and a modified AHEI score (excluding overlapping components), a higher PDS was significantly associated with a lower risk of total CVD, CHD, and stroke. The associations did not change when further adjusting for potential mediators, including baseline hypercholesterolemia, hypertension, and diabetes, therefore only the pooled model with these covariates is shown (Table 2). In the pooled analyses of the three cohorts, the fully adjusted HRs (95% CIs) for total CVD comparing the highest quintile to the lowest quintile of the PDS was 0.86 (0.81 to 0.92; p-trend <0.001). The pooled HRs (95% CIs) for CHD was 0.86 (0.80 to 0.93; p-trend=0.0001) and for stroke was 0.86 (0.78 to 0.95; p-trend=0.0003), comparing extreme quintiles. The estimated absolute risk reduction (ARR) and number needed to treat (NNT) when comparing extreme quintiles was 1.1% (NNT=91) for total CVD, which estimates how many participants need to improve their diet score from the bottom to the top quintile to prevent one cardiovascular event.

Table 2:

Association of the Portfolio diet (cumulative average) with cardiovascular disease outcomes in 73,924 women from the NHS (1984–2016), 92,346 women from the NHS II (1991–2017) and 43,970 men from the HPFS (1986–2016).

	Q1 HR (95% CIs)	Q2 HR (95% CIs)	Q3 HR (95% CIs)	Q4 HR (95% CIs)	Q5 HR (95% CIs)	P value for trend	HR (95% CI) for 25-percentile increment (6 points)
TOTAL CARDIOVASCULAR DISEASE
NHS
Cases/person-years	1718/407,810	1692/406,649	1589/439,795	1460/392,886	1541/413,058
Age-adjusted Model 1	1.00 [reference]	0.92 (0.86, 0.99)	0.82 (0.77, 0.89)	0.78 (0.73, 0.84)	0.76 (0.71, 0.81)	<0.001	0.84 (0.81, 0.87)
Multivariable Model 2	1.00 [reference]	1.00 (0.94, 1.07)	0.93 (0.86, 1.00)	0.92 (0.85, 1.00)	0.91 (0.83, 0.99)	0.0002	0.93 (0.89, 0.97)
NHSII
Cases/person-years	439/513,711	341/487,152	356/511,636	346/516,677	277/493,995
Age-adjusted Model 1	1.00 [reference]	0.77 (0.67, 0.88)	0.80 (0.70, 0.92)	0.73 (0.63, 0.84)	0.57 (0.49, 0.67)	<0.001	0.76 (0.70, 0.81)
Multivariable Model 2	1.00 [reference]	0.82 (0.71, 0.95)	0.88 (0.75, 1.02)	0.80 (0.68, 0.95)	0.67 (0.55, 0.81)	0.0002	0.81 (0.74, 0.89)
HPFS
Cases/person-years	1448/208,244	1412/214,425	1393/227,485	1434/206,341	1471/220,459
Age-adjusted Model 1	1.00 [reference]	0.92 (0.86, 0.99)	0.82 (0.77, 0.89)	0.86 (0.79, 0.92)	0.79 (0.73, 0.85)	<0.001	0.89 (0.86, 0.92)
Multivariable Model 2	1.00 [reference]	0.97 (0.90, 1.05)	0.89 (0.82, 0.96)	0.93 (0.86,1.01)	0.89 (0.81, 0.97)	0.006	0.95 (0.90, 0.99)
Pooled model 2	1.00 [reference]	0.97 (0.92, 1.01)	0.90 (0.86, 0.95)	0.90 (0.86, 0.95)	0.86 (0.81, 0.92)	<0.001	0.92 (0.89, 0.95)
CORONARY HEART DISEASE
NHS
Cases/person-years	964/408,265	916/407,162	823/440,332	802/393,315	825/413,530
Age-adjusted Model 1	1.00 [reference]	0.89 (0.81, 0.98)	0.75 (0.69, 0.83)	0.77 (0.70, 0.84)	0.72 (0.66, 0.79)	<0.001	0.81 (0.77, 0.85)
Multivariable Model 2	1.00 [reference]	0.99 (0.91, 1.09)	0.89 (0.80, 0.98)	0.95 (0.85, 1.06)	0.92 (0.82, 1.04)	0.13	0.92 (0.86, 0.98)
NHSII
Cases/person-years	239/513,925	162/487,324	184/511,816	191/516,816	134/494,127
Age-adjusted Model 1	1.00 [reference]	0.67 (0.55, 0.81)	0.76 (0.63, 0.92)	0.73 (0.61, 0.89)	0.51 (0.41, 0.62)	<0.001	0.73 (0.66, 0.81)
Multivariable Model 2	1.00 [reference]	0.77 (0.63, 0.95)	0.94 (0.76, 1.15)	0.94 (0.75, 1.17)	0.72 (0.55, 0.94)	0.11	0.88 (0.76, 1.00)
HPFS
Cases/person-years	1115/208,555	1080/214,730	1056/227,784	1088/206,639	1087/220,798
Age-adjusted Model 1	1.00 [reference]	0.91 (0.84, 0.99)	0.81 (0.74, 0.88)	0.84 (0.77, 0.92)	0.75 (0.69, 0.82)	<0.001	0.87 (0.83, 0.90)
Multivariable Model 2	1.00 [reference]	0.97 (0.89, 1.06)	0.89 (0.81, 0.97)	0.93 (0.85, 1.02)	0.86 (0.77, 0.95)	0.003	0.93 (0.88, 0.98)
Pooled model 2	1.00 [reference]	0.96 (0.90, 1.02)	0.88 (0.83, 0.94)	0.93 (0.87, 0.99)	0.86 (0.80, 0.93)	0.0001	0.92 (0.88, 0.95)
TOTAL STROKE
NHS
Cases/person-years	808/408,269	819/407,155	800/440,243	694/393,355	764/413,492
Age-adjusted Model 1	1.00 [reference]	0.95 (0.86, 1.05)	0.88 (0.79, 0.97)	0.79 (0.71, 0.87)	0.79 (0.72, 0.88)	<0.001	0.87 (0.82, 0.91)
Multivariable Model 2	1.00 [reference]	1.00 (0.90, 1.10)	0.95 (0.85, 1.06)	0.87 (0.77, 0.98)	0.88 (0.79, 1.00)	0.02	0.93 (0.87, 1.00)
NHSII
Cases/person-years	201/513,915	182/487,291	172/511,803	158/516,838	143/494,121
Age-adjusted Model 1	1.00 [reference]	0.90 (0.73, 1.10)	0.84 (0.69, 1.04)	0.74 (0.60, 0.91)	0.66 (0.53, 0.81)	<0.001	0.78 (0.71, 0.87)
Multivariable Model 2	1.00 [reference]	0.88 (0.72, 1.09)	0.82 (0.65, 1.01)	0.69 (0.54, 0.88)	0.62 (0.47, 0.82)	0.0002	0.75 (0.68, 0.86)
HPFS
Cases/person-years	333/208,806	332/214,961	337/228,071	346/206,894	384/220,987
Age-adjusted Model 1	1.00 [reference]	0.96 (0.82, 1.11)	0.88 (0.76, 1.02)	0.90 (0.77, 1.04)	0.90 (0.77, 1.05)	0.13	0.95 (0.88, 1.02)
Multivariable Model 2	1.00 [reference]	0.98 (0.84, 1.14)	0.91 (0.78, 1.07)	0.94 (0.80, 1.12)	0.97 (0.81, 1.17)	0.68	0.99 (0.90, 1.08)
Pooled model 2	1.00 [reference]	0.97 (0.90, 1.05)	0.92 (0.85, 1.00)	0.85 (0.78, 0.93)	0.86 (0.78, 0.95)	0.0003	0.92 (0.87, 0.96)

The portfolio diet score was calculated based on cumulatively averaged dietary intakes from all preceding food frequency questionnaires up to each follow-up interval. Q5 reflects highest adherence the diet, Q1 reflects lowest. Multivariable model 1 was adjusted for age (months). Model 2 was adjusted for age and race (White, non-White), smoking (never; past; current: 1–14, 15–24, ≥ 25 cigarettes/day), menopausal status and post-menopausal hormone use (premenopausal, never/past users of hormone therapy, and current users of hormone therapy; only in NHS and NHSII), oral contraceptive use (never, past, and current; only in NHSII), multivitamin use (no, yes), regular aspirin use (no, yes), physical activity (<3, 3–9, 9–18, 18–27, 27–42, ≥42 metabolic equivalents/week), family history of myocardial infarction (no, yes), family history of diabetes (no, yes), marital status (married, widowed, divorced/separated), body mass index (<23, 23–24.9, 25–29,9, 30–34.9, ≥35 kg/m²), alcohol intake, total energy intake, and modified AHEI (in quintiles) and baseline hypercholesterolemia (no, yes), hypertension (no, yes) and diabetes (no, yes). Results were pooled using a pooled dataset. CI=confidence interval; HPFS=Health Professionals Follow-up Study; HR=hazard ratio; NHS=Nurses’ Health Study; Q=quintile.

In addition, a 25-percentile higher PDS was associated with a lower risk of total CVD (pooled HR: 0.92; 95% CIs: 0.89 to 0.95), CHD (pooled HR: 0.92; 95% CI: 0.88 to 0.95) and stroke (pooled HR: 0.92; 95% CI: 0.87 to 0.96) (Table 2). In dose-response analyses, there was no evidence of departure from linearity for total CVD, CHD or stroke (tests for curvature: p=0.10 for CVD, p=0.11 for CHD and p=0.27 for stroke) (Figure 1). In the stroke subtype analysis, the pooled HR (95% CIs) for the three cohorts was 0.93 (0.81 to 1.08, p-trend=0.15) for ischemic stroke and 0.84 (0.64 to 1.09; p-trend=0.24) for hemorrhagic stroke. In addition, the HR (95% CIs) was 0.92 (0.85 to 0.99) and 1.00 (0.87 to 1.14) for a 25-percentile higher PDS for ischemic and hemorrhagic stroke, respectively (Table S3).

Figure 1: Dose-Response Relationship of the Portfolio Diet Score with Risk of CVD, CHD, and Stroke.

(A) CVD, (B) CHD, (C) Stroke. Analysis conducted after combining all three cohorts. Adjusted for age, race, smoking, menopausal status and post-menopausal hormone use (only in NHS and NHSII), oral contraceptive use (only in NHSII), multivitamin use, regular aspirin use, physical activity, family history of myocardial infarction, family history of diabetes, marital status, body mass index, total energy intake, modified AHEI, alcohol intake, hypercholesterolemia, hypertension and diabetes at baseline. Solid lines represent hazard ratios and dotted lines represent 95% confidence intervals.

Sensitivity and Subgroup Analyses:

In the sensitivity analyses, results remained largely consistent comparing highest to lowest PDS quintiles. The HRs (95% CIs) for total CVD, CHD and stroke when further adjusted for SES were 0.86 (0.81 to 0.91), 0.85 (0.79 to 0.92) and 0.86 (0.78 to 0.95) when comparing extreme quintiles, respectively. When adjusting for updated diagnoses of hypercholesterolemia or medication, hypertension or medication, and diabetes, the HR (95% CIs) was 0.86 (0.81 to 0.91) for total CVD, 0.85 (0.79 to 0.92) for CHD and 0.86 (0.78 to 0.95) for stroke. Using the most recent measure of diet, the HR (95% CIs) was 0.86 (0.81 to 0.90) for total CVD, 0.82 (0.77 to 0.87) for CHD and 0.91 (0.84 to 0.99) for stroke. Lastly, continuously updating diet until the end of follow-up resulted in HRs (95% CIs) of 0.86 (0.81 to 0.91) for total CVD, 0.86 (0.79 to 0.92) for CHD and 0.85 (0.77 to 0.94) for stroke.

In the stratified analyses, the associations were consistent across most subgroups and the risk of CVD. However, the association between the PDS and CVD risk was significantly stronger among participants younger than 60 years of age (p-difference<0.001), those who were more physically active (p-difference=0.02), and ever-smokers (p-difference<0.001) (Table 3). These findings were consistent for CHD (Table S4), however, only the interaction for a stronger association in those younger than 60 years of age was significant for stroke (Table S5). No significant interactions were observed for BMI, sex, family history of MI, hypertension status, hypercholesterolemia status and racial/ethnic groups for CVD, CHD, and stroke. Lastly, comparing extreme quintiles, BMI did not mediate the PDS relationship with total CVD or stroke, and with CHD, BMI explained 4.3% of the total relation (p=0.046).

Table 3:

Subgroup Analysis for Risk of Total CVD According to the Portfolio Diet Score

	NHS HR (95% CI)	P interaction	NHS II HR (95% CI)	P interaction	HPFS HR (95% CI)	P interaction	Pooled HR (95% CI)	P interaction
Age, yrs		0.003		0.34		0.25		<0.001
<60	0.82 (0.72, 0.92)		0.81 (0.72, 0.90)		0.90 (0.80, 1.01)		0.82 (0.77, 0.88)
≥60	0.95 (0.90, 0.99)		0.83 (0.68, 1.02)		0.95 (0.91, 1.00)		0.94 (0.91, 0.97)
BMI, kg/m2		0.83		0.86		0.68		0.98
<30	0.92 (0.87, 0.97)		0.79 (0.69, 0.89)		0.93 (0.89, 0.98)		0.91 (0.87, 0.94)
≥30	0.95 (0.86, 1.05)		0.84 (0.72, 0.98)		0.98 (0.86, 1.10)		0.92 (0.86, 0.99)
Sex		--		--		--		0.18
Women	0.93 (0.89, 0.97)		0.81 (0.74, 0.89)		--		0.90 (0.86, 0.94)
Men	--		--		0.95 (0.90, 0.99)		0.95 (0.90, 0.99)
Family history of MI		0.71		0.48		0.02		0.09
Yes	0.94 (0.87, 1.02)		0.82 (0.69, 0.98)		0.90 (0.80, 1.00)		0.89 (0.83, 0.94)
No	0.94 (0.88, 0.99)		0.84 (0.74, 0.94)		0.97 (0.92, 1.01)		0.92 (0.89, 0.95)
Physical activity		0.55		0.98		0.001		0.02
Below median	0.91 (0.86, 0.97)		0.84 (0.74, 0.96)		1.01 (0.94, 1.08)		0.93 (0.89, 0.96)
Above median	0.94 (0.87, 1.01)		0.74 (0.64, 0.86)		0.89 (0.84, 0.94)		0.88 (0.84, 0.92)
Smoking		<0.001		0.02		0.47		<0.001
Ever smoker	0.85 (0.80, 0.91)		0.71 (0.61, 0.82)		0.95 (0.89, 1.00)		0.87 (0.83, 0.90)
Never smoker	0.98 (0.91, 1.06)		0.85 (0.75, 0.98)		0.91 (0.85, 0.98)		0.92 (0.88, 0.96)
Hypertension		0.89		0.88		0.97		0.83
Yes	0.91 (0.84, 0.99)		0.93 (0.74, 1.16)		0.98 (0.91, 1.06)		0.94 (0.89, 0.99)
No	0.93 (0.88, 0.99)		0.79 (0.70, 0.88)		0.93 (0.88, 0.98)		0.90 (0.87, 0.94)
Hypercholesterolemia		0.70		0.91		0.32		0.53
Yes	0.98 (0.85, 1.14)		0.82 (0.67, 0.99)		0.94 (0.83, 1.07)		0.93 (0.85, 1.01)
No	0.92 (0.87, 0.97)		0.81 (0.73, 0.91)		0.95 (0.90, 0.99)		0.91 (0.87, 0.94)
Racial/ethnic groups *		0.28		0.11		0.85		0.39
Non-Hispanic white	0.94 (0.89, 0.98)		0.85 (0.77, 0.94)		0.96 (0.92, 1.00)		0.92 (0.89, 0.95)
Minority racial/ethnic group	1.03 (0.74, 1.43)		0.33 (0.19, 0.59)		0.82 (0.66, 1.02)		0.78 (0.66, 0.93)

HRs are for 25% increment increase in the portfolio diet score (6 points) in each subgroup category. Multivariable model adjusted for age (months), race (White, non-White), smoking (never; past; current: 1–14, 15–24, ≥ 25 cigarettes/day), menopausal status and post-menopausal hormone use (premenopausal, never/past users of hormone therapy, and current users of hormone therapy; only in NHS and NHSII), oral contraceptive use (never, past, and current; only in NHSII), multivitamin use (no, yes), regular aspirin use (no, yes), physical activity (<3, 3–9, 9–18, 18–27, 27–42, ≥42 metabolic equivalents/week), family history of myocardial infarction (no, yes), family history of diabetes (no, yes), marital status (married, widowed, divorced/separated), body mass index (<23, 23–24.9, 25–29,9, 30–34.9, ≥35 kg/m²), alcohol intake, total energy intake and modified AHEI (in quintiles), and baseline hypercholesterolemia (no, yes), hypertension (no, yes) and diabetes (no, yes). HPFS=Health Professionals Follow-up Study; HR=hazard ratio; MI=myocardial infarction; NHS=Nurses’ Health Study

^*Non-Hispanic white includes non-Hispanic individuals with southern European/Mediterranean ancestry, Scandinavian ancestry, and other Caucasian ancestry. Minority racial/ethnic groups include non-Hispanic black, Hispanic, and other racial/ethnic groups. Other includes those not classified as non-Hispanic white, non-Hispanic black, or Hispanic, such as Asian and American Indian.

Biomarker Analysis:

In the plasma biomarker analyses, cross-sectional associations of the PDS were examined with levels of lipid and inflammatory biomarkers after adjustment for covariates, including BMI. An inverse association between higher PDS quintiles and Total-C (p trend=0.0003) and non-HDL-C (p trend=0.01) was observed. The p for trends were not significant for LDL-C, HDL-C and triglycerides, however, some quintiles were associated with lower triglycerides levels and the pattern for LDL-C was trending downwards with higher PDS quintiles (Figure 2). Further, a significant inverse association between higher PDS quintiles and IL-6 (p trend=0.0002), hsCRP (p trend<0.001) and leptin (p trend=0.002) was found. The p trends were not significant for sICAM-1, TNFα-R1, TNFα-R2, and adiponectin, however, some higher quintiles of the PDS were associated with higher adiponectin levels (Figure 3).

Figure 2: Association between the portfolio diet score and levels of lipid biomarkers.

Linear regressions were used to analyze associations between average of two PDS (average of 1984 and 1986 in NHS, 1991 and 1995 in NHSII and 1986 and 1990 in HPFS). Multivariable models were adjusted for study cohort, age, fasting status, BMI, race, smoking, aspirin use, other anti-inflammatory medications, multivitamin use, menopausal status and post-menopausal hormone use (in women), physical activity, modified AHEI, energy intake, alcohol intake, hypercholesterolemia, hypertension, diabetes, family history of CVD, and case-control status in original substudies. Squares represent SD differences in biomarkers comparing higher to the lowest PDS quintiles, and vertical lines represent 95% CIs. HDL-C=high-density lipoprotein cholesterol; LDL-C=low-density lipoprotein cholesterol; non-HDL-C=non-high-density lipoprotein cholesterol; PDS=portfolio diet score; Total-C=total cholesterol.

Figure 3: Association between the portfolio diet score and levels of inflammatory biomarkers.

Linear regressions were used to analyze associations between the average of two PDS (average of 1984 and 1986 in NHS, 1991 and 1995 in NHSII and 1986 and 1990 in HPFS). Multivariable models were adjusted for study cohort, age, fasting status, BMI, race, smoking, aspirin use, other anti-inflammatory medications, multivitamin use, menopausal status and post-menopausal hormone use (in women), physical activity, modified AHEI, energy intake, alcohol intake, hypercholesterolemia, hypertension, diabetes, family history of CVD, and case-control status in original substudies. Squares represent SD differences in biomarkers comparing higher to the lowest PDS quintiles, and vertical lines represent 95% CIs. hsCRP=high sensitivity C-reactive protein; PDS=portfolio diet score; sICAM-1=soluble intercellular adhesion molecule-1; TNFα-R1=tumor necrosis factor-α receptor 1; TNFα-R2=tumor necrosis factor-α receptor 2.

Individual PDS Component Analysis:

Lastly, quintiles of the individual components of the PDS (in servings/day, except for phytosterols which were in mg/day) with risk of total CVD, CHD and stroke were analyzed. For total CVD and CHD, higher intake of viscous fiber sources, nuts, MUFAs, and phytosterols were associated with lower risk and higher intake of saturated fat and cholesterol sources was associated with a higher risk. For stroke, only higher intake of viscous fiber sources and phytosterols were associated with a lower risk and higher intake of saturated fat and cholesterol sources was associated with a higher risk. However, several of these associations were attenuated and became non-significant after mutual adjustment for the other PDS components (Table S6).

DISCUSSION

In these three large prospective cohort studies, the PDS was associated with a 14% lower risk of total CVD, CHD and stroke, when comparing extreme quintiles, with no evidence of departure from linearity. These findings remained consistent across several sensitivity analyses and most subgroups, and after adjustment for the AHEI (excluding overlapping components). A higher PDS was also associated with a more favorable lipid profile and lower levels of inflammation. No associations were observed between the PDS and stroke subtypes when comparing extreme quintiles, however, the 25-percentile increase in the PDS was associated with a lower risk of ischemic stroke.

Comparison with previous literature

These findings are consistent with prior research assessing the PDS with CVD outcomes. The Women’s Health Initiative (WHI) prospective cohort study demonstrated that a higher PDS was associated with an 11% lower risk of total CVD and a 14% lower risk of CHD¹⁰, which are similar to the main pooled findings of the current study, as well as in the stratification analysis by women only. However, contrary to the current findings, the association with stroke was not significant in the WHI. This study, however, had a larger number of stroke cases (n=6,473 versus n=4,440 in the WHI), therefore it may have been better powered to detect an association. Additionally, the inverse association with stroke was stronger among younger participants, whereas the WHI included postmenopausal women. In this study, the associations of the PDS with stroke subtypes were not significant when comparing extreme quintiles, although a trend towards a lower risk of both ischemic and hemorrhagic stroke in the pooled analyses was observed. Although the portfolio dietary pattern may not be exclusively vegetarian, previous studies on vegetarian diets and hemorrhagic stroke risk have shown conflicting results, with some showing a lower risk²² and others a higher risk²³. Further research is needed to understand the potential role of the portfolio dietary pattern in reducing the risk of total and subtypes of stroke. The finding of stronger associations in younger adults also warrants further investigation to better understand the role of the diet in older adults.

These findings are also in agreement with other dietary patterns frequently recommended for CVD risk reduction. These include the Mediterranean, DASH, Nordic, and healthy vegetarian and plant-based dietary patterns, all of which share core foods (nuts, legumes, whole grains, fruits, vegetables, and healthy unsaturated plant oils), and have also been associated with lower risk of CVD in prospective cohort studies^24–28. Also similar to these findings, the Healthy Eating Index-2015, aMED, hPDI and AHEI dietary patterns were all associated with a 14–21% lower risk of CVD when comparing highest versus lowest quintiles, including both CHD and stroke, in the NHS, NHSII and HPFS cohorts²⁹. Furthermore, a study of the plant-based diet indices in the same cohorts found that the hPDI was associated with a 10% lower risk of total stroke when comparing the extreme quintiles³⁰. Likewise, no significant associations were found with either ischemic or hemorrhagic stroke. The lack of significant associations and wider CIs seen with the extreme quintiles of the plant-based diet indices and the PDS in the NHS, NHSII, and HPFS cohorts may be attributed to fewer cases and limited precision when analyzing stroke subtypes. Additionally, the correlation analysis showed that the PDS was most highly correlated with the Mediterranean diet, consistent with previous research that examined the portfolio, DASH, and Mediterranean diets with incident type 2 diabetes in the WHI³¹. Beyond prospective cohort studies, the Mediterranean diet has been examined with primary prevention of clinical CVD events in an RCT. The Prevención con Dieta Mediterránea (PREDIMED) trial examined the effect of a Mediterranean diet supplemented with either extra virgin olive oil or nuts compared with a low-fat diet and found reductions in major cardiovascular events of 31% and 28%, respectively³². However, none of the dietary pattern indices were strongly correlated with the PDS, indicating that each diet consists of a unique combination of foods and nutrients and that multiple dietary patterns can be recommended for CVD prevention based on patient preferences. The dietary components that differentiate the PDS from other diet quality indices include a higher emphasis on plant protein, particularly soy, viscous fiber sources and phytosterols.

The inverse associations of the PDS with CVD are also in line with the evidence from RCTs on intermediate risk factors for CVD. The portfolio diet has shown clinically meaningful reductions in the primary lipid targets for CVD prevention (LDL-C, non-HDL-C, and ApoB), as well as reductions in CRP, Total-C, triglycerides, and blood pressure⁶. In the cross-sectional analyses of lipid and inflammatory biomarkers, higher quintiles of the PDS were linearly associated with lower Total-C and non-HDL-C, with some PDS quintiles being associated with lower triglyceride levels. The association with LDL-C was not significant, although there was a trend for lower levels with increasing PDS quintiles. LDL-C was the main target of the portfolio dietary pattern in previous RCTs and these studies have shown reductions in LDL-C up to ~30%^{6, 9}, however, previous observational research has reported inconsistent findings regarding the association between the PDS and LDL-C. While the previous validation study found an association between PDS and lower levels of LDL-C¹⁶, it measured changes in LDL-C levels rather than a cross-sectional analysis like the current study. However, another study did not observe significant associations between the PDS and changes in LDL-C³³, and previous analyses of the empirical dietary inflammatory pattern and olive oil consumption in the NHS and HPFS cohorts also did not observe significant associations with LDL-C, but did find associations with other biomarkers assessed in this study^{34, 35}. Significant inverse associations were, however, observed with non-HDL-C. Non-HDL-C represents all atherogenic particles and is now recognized as a better marker for CVD risk assessment^36–38. Thus, although the associations with LDL-C were not significant in the current study, the inverse associations seen with non-HDL-C suggest that the reduction of atherogenic blood lipids may be a potential mechanism for the observed association between the PDS and lower CVD risk.

The association with lower hsCRP in this study is also consistent with the RCT evidence where the portfolio dietary pattern reduced hsCRP levels by ~30%⁶. The associations with other inflammatory markers, IL-6, leptin and adiponectin, have not been previously assessed with the portfolio diet, and offer other potential mechanisms in which the diet may lower CVD risk. The inverse association with IL-6 and leptin, and positive association with adiponectin, have been observed in previous studies of plant-based dietary patterns. For example, the hPDI was similarly associated with these inflammatory and metabolic biomarkers in an analysis of women in the NHS³⁹. Moreover, a systematic review and meta-analysis of RCTs found that plant-based diets improved inflammatory profiles, with marked reductions in CRP, IL-6 and sICAM⁴⁰, even after controlling for weight loss. However, unlike the current study findings, plant-based diets did not affect adiponectin and leptin levels in RCTs. Inflammation, as reflected by changes in hsCRP levels, may play an important role in the pathogenesis of CVD^{41, 42}. Additionally, adiponectin exerts anti-inflammatory and anti-atherosclerotic properties⁴³ and leptin has been implicated in the development of CHD in patients with type 2 diabetes⁴⁴. The role of the portfolio dietary pattern in preventing cardiometabolic disease through improvements in these biomarkers warrants further investigation as they may provide insight into other biological mechanisms for reducing CVD risk beyond the lipid and hsCRP-lowering properties of the diet previously tested in RCTs.

Strengths and Limitations

This study has several strengths, including the prospective cohort design with long-term follow-up, the large size of the cohorts, high rate of follow-up, and repeated dietary measurements over three decades, which helps reduce measurement errors while better reflecting long-term dietary patterns. The biomarker analyses also complemented the prospective cohort analyses of diet and clinical CVD outcomes. Limitations include the observational study design, therefore residual confounding cannot be ruled out, although several repeated measurements of potential confounders were controlled for. Diet was also self-reported which may result in measurement errors, however, the FFQs are validated with diet records and biomarkers^13–15. Baseline LDL-C was only measured in a subset of participants, which prevented additional analyses of LDL-C levels in the cohorts. Lastly, consumption of some foods recommended in the portfolio diet remained low, even in the top quintiles, and therefore the association with CVD risk may be underestimated. A stronger association may be seen with greater consumption of the portfolio diet foods such as soy products, beans, and viscous fiber sources. Additionally, information on a few key portfolio diet foods was not available, including some viscous fiber sources (barley, okra) and phytosterol supplements and fortified foods, potentially leading to underestimation of intake and misclassification of some participants, which may also attenuate the findings. However, consumption of these foods is likely to be low in the US population.

Implications

This study provides additional evidence to support the use of the plant-based portfolio dietary pattern for reducing the risk of CVD, as highlighted by several cardiovascular clinical practice guidelines globally^{45, 46}. This dietary pattern also aligns with the American Heart Association’s guidelines promoting the consumption of whole grains, fruits and vegetables, healthy plant-based proteins, minimally processed foods, and healthy unsaturated plant oils⁴⁷. Moreover, many of the recommended foods in the portfolio diet have a low environmental impact⁴⁸, making it a beneficial option for both personal and planetary health. The finding that some portfolio diet foods were not highly consumed also represents an opportunity for individuals to achieve the cardiovascular benefits of the diet, as well as the public health need to create an environment that facilitates higher intake of these heart-healthy foods. However, despite the lower consumption of some foods, these findings highlight that even partial adoption of the portfolio dietary pattern can confer cardiovascular benefits. For example, comparing extreme quintiles of the PDS was associated with a 14% lower incidence of CVD, even though the consumption of portfolio diet foods in the top quintiles of the cohorts was lower than the recommendations from the RCTs. Practically, to achieve the intake in the highest quintile, one could add approximately ½ cup of beans or 1 cup of soymilk, 1 cup of cooked oatmeal and ½ cup eggplant, 0.5 to 1 ounce of nuts, and 1 tbsp of olive oil to their diets per day through substitutions for foods higher in saturated fat and cholesterol, such as red and processed meat, butter, eggs with yolk and high fat dairy.

CONCLUSIONS

Findings from these three large prospective cohort studies showed that the PDS was significantly associated with a lower risk of total CVD, CHD and stroke. The PDS was also inversely associated with several blood lipids and hsCRP. The associations with other inflammatory biomarkers, including IL-6 and leptin, deserve further investigation as potential mechanisms linking this dietary pattern with lower CVD risk. Confirming these findings in other populations and in RCTs with incident CVD events is also warranted. Overall, these results support clinical practice and dietary guidelines that emphasize the consumption of cholesterol-lowering plant-based foods for CVD prevention.

CLINICAL PERSPECTIVE.

What is new?

• The plant-based portfolio diet score of established cholesterol-lowering foods was linearly and consistently associated with a 14% lower risk of cardiovascular disease (CVD), coronary heart disease and stroke in three prospective cohort studies

• The portfolio diet score was also associated with a more favorable blood lipid and inflammatory profile

What are the clinical implications?

• These findings support the use of the portfolio dietary pattern for reducing the risk of CVD, consistent with several cardiovascular clinical practice guidelines globally

• This study also highlights that even partial adoption of the portfolio dietary pattern can confer cardiovascular benefits

NONSTANDARD ABBREVIATIONS AND ACRONYMS

AHEI

alternate healthy eating index

ApoB

apolipoprotein B

aMED

alternate Mediterranean diet

ARR

absolute risk reduction

CHD

coronary heart disease

CI

confidence interval

CVD

cardiovascular disease

DASH

dietary approaches to stop hypertension

FFQ

food frequency questionnaire

HDL-C

high-density lipoprotein cholesterol

hPDI

healthy plant-based diet index

HPFS

Health Professionals Follow-up Study

HR

hazard ratio

hsCRP

high-sensitivity C-reactive protein

IL-6

interleukin-6

LDL-C

low-density lipoprotein cholesterol

MI

myocardial infarction

MUFA

monounsaturated fat

NHS

Nurses’ Health Study

NNT

number needed to treat

Non-HDL-C

non-high-density lipoprotein cholesterol

PDI

plant-based diet index

PDS

portfolio diet score

SES

socioeconomic status

sICAM-1

soluble intercellular adhesion molecule-1

TNFα-RI

tumor necrosis factor-α receptor 1

TNFα-R2

tumor necrosis factor-α receptor 2

Total-C

total cholesterol

uPDI

unhealthy plant-based diet index

WHI

Women’s Health Initiative