Specific dietary protein sources are associated with cardiometabolic risk factors in the Boston Puerto Rican Health Study

Abstract

Background:

Puerto Rican adults residing in the US mainland experience a high prevalence of metabolic syndrome (MetS). A diet containing healthy protein-rich sources may help control risk factors comprising MetS.

Objective:

This study aimed to evaluate 2-year longitudinal associations between intake of various protein-rich foods and changes in the six MetS components.

Design:

This is a secondary analysis of a longitudinal cohort study using data from the baseline (2004–2007) and 2-year follow-up visits (2006–2011) in the Boston Puerto Rican Health Study. Participants/setting: Participants were self-identified Puerto Ricans, aged 45–75 y, residing in Boston, Massachusetts or the surrounding area (n=1126).

Main outcome measures:

MetS components were fasting glucose, HDL cholesterol, triglycerides, systolic and diastolic blood pressures, and waist circumference.

Statistical analysis:

Baseline intake of foods reported in a semi-quantitative food frequency questionnaire were expressed as servings/day, and protein-rich foods were categorized as: unprocessed white meat, unprocessed red meat, processed meat, milk and yogurt, cheese, fish and seafood, beans, nuts, and eggs. Associations between each continuous protein food group and continuous 2-year change in MetS components were assessed using linear mixed models adjusted for socioeconomic and behavioral factors, and other dietary sources.

Results:

The top contributors to total protein intake were unprocessed red meat (13.3%) and unprocessed poultry (13.0%), and the lowest were eggs (2.92%) and nuts (0.91%). Higher intake of processed meats was associated with an increase in waist circumference over 2 years (β=1.28; SE=0.63), whereas higher intake of fish and seafood was associated with a decrease in waist circumference (β=−3.47; SE=1.39). Intake of unprocessed poultry was associated with a decrease in triglycerides (β=−24.5; SE=9.13). No other significant associations were observed between protein sources and two-year changes in MetS components.

Conclusions:

Consuming less processed meat and more fish and seafood and unprocessed poultry was associated with decreases in waist circumference and triglycerides among US mainland Puerto Ricans. Other dietary protein sources were not related to cardiometabolic health.

Introduction

Metabolic syndrome (MetS) encompasses various risk factors for cardiovascular disease and type 2 diabetes (T2D) and is of increasing concern in the United States (US).^1,2 These risk factors include high waist circumference, dyslipidemia, hyperglycemia, and hypertension.² Puerto Rican adults residing in the US mainland experience profound health disparities in prevalence of MetS.³ In a 2008–2011 study among Hispanic/Latino individuals living in the US, Puerto Ricans had the highest age-standardized prevalence of MetS at 37.1% (95% CI: [34.3, 39.9]).⁴ Compared to non-Latino whites and other Latino heritages, Puerto Ricans experience some of the highest prevalence of obesity, T2D⁵, hypertension, and hyperlipidemia.⁶ Because of this, it is imperative to understand modifiable risk factors influencing MetS in this population.

Consuming a protein-rich diet may help alleviate clinical outcomes related to MetS.^7–9 For example, a protein rich diet has been shown to improve weight loss and weight maintenance, as well as blood pressure and blood lipids.¹⁰ However, protein sources are important to consider for reduction of risk as they differ in their amino acid makeup, non-protein nutrient content, and digestibility. High dietary intake of unprocessed and processed red meat has been associated with higher risk of both obesity and T2D.^11–14 Results are mixed when processed and unprocessed red meat are assessed separately.¹⁵ In contrast, protein intake from vegetarian sources (i.e., nuts and seeds) has shown protective effects against MetS risk.^16–19 However, the literature does not show consistent results for other dietary protein sources. While beef, chicken, and dairy have been shown to negatively affect blood lipid concentrations,²⁰ other studies have found that intake of dairy was associated with decreased risk of T2D, obesity, and hypertension.^12,21,22 There have been inconsistent or inconclusive results in the literature surrounding other dietary protein sources, such as eggs and fish.^23–26 Moreover, many studies have been conducted in predominantly non-Latino white populations, and dietary patterns of the populations and dietary assessment tools across studies are heterogeneous. Thus, it is important to understand how diverse protein sources may influence MetS risk factors in Puerto Ricans, as their cultural food preferences differ from the populations from which most of the evidence has been gathered so far.

Puerto Ricans consume a wide variety of protein-rich foods. In the Boston Puerto Rican Osteoporosis Study, the top five foods contributing to protein intake were beans, chicken legs with skin, white rice, chicken legs without skin, and whole milk.²⁷ A study of adults living in Puerto Rico found that only 1.2% of the participants adhered to the daily recommended amount of less than 1.5 servings/day of red/processed meats, and that the mean amount of animal protein consumed daily was more than 80% higher than vegetable protein (51% vs. 27%).²⁸ Puerto Ricans living in the US have been found to have worse overall dietary intake and lower adherence to recommended intake of foods in the Alternate Healthy Eating Index (AHEI), compared with other US Hispanic groups.²⁹ Additionally, Puerto Ricans have been found to have the strongest association among US Hispanics between AHEI and decreased waist circumference, as well as significant associations with decreased diastolic blood pressure and fasting plasma glucose, and an association with increased HDL cholesterol.³⁰ Thus, given the high prevalence of MetS in Puerto Ricans and the culturally-specific protein-rich foods consumed by this group that may distinctively influence cardiometabolic risk factors, this study aimed to evaluate longitudinal associations between intake of diverse protein-rich foods and changes in MetS components.

Methods

Study Population

This secondary analysis used data from the Boston Puerto Rican Health Study, a longitudinal cohort study of 1500 self-identified Puerto Ricans aged 45–75 years old residing in Boston, Massachusetts or the surrounding area. The methods for this study have been described in detail elsewhere.³¹ Briefly, the study consists of two data collection time points: baseline (2004–2007) and ~2-year follow-up (2006–2011). Participants were recruited through door-to-door enumeration in highly Hispanic-populated blocks identified through the 2000 Census data, and supplemental recruitment was conducted within the community. One eligible participant per household was invited to participate in the study. Participants with severe health conditions or cognitive limitation, defined as Mini-Mental State Examination score ≤10, and participants who planned to move away from the area within two years were excluded from the study. Participants provided written informed consent prior to study participation. The study was approved by the Institutional Review Boards of Tufts and Northeastern Universities. No prior sample size estimation was performed for the current analysis.

The current analysis used baseline and 2-year data. A total of 1,221 participants completed the 2-year follow-up. Participants were excluded if they had implausible dietary intakes (<600 or > 4,800 kcal/day; n=56), incomplete FFQ data (n=23), or missing covariate data (n=16). Thus, the maximum sample size for each model was 1,126, although the final sample sizes vary based on missing data for each outcome and food group.

Data Collection and Measures

All data were collected in the homes of the participants. Trained bilingual interviewers administered a questionnaire to collect data on participant demographics, medical and health-behavior history, and medication use.

Dietary Protein Sources

A semi-quantitative food frequency questionnaire (FFQ) was used to collect data on dietary intake over the past 12 months at the baseline time point. The FFQ was adapted for the population by including common foods, portion sizes, and recipes typical for Puerto Ricans.³² The FFQ has been validated against vitamins E and B12 and plasma carotenoids among Hispanics aged 60 years and older.^33–35 Analysis of dietary intake was conducted using the Nutrition Data System for Research software version 2007 developed by the Nutrition Coordinating Center (NCC), University of Minnesota, Minneapolis, MN.³⁶

Reported intakes of foods at baseline were expressed as servings per day, and protein-rich foods were categorized into nine groups: unprocessed white meat (e.g. chicken, turkey), unprocessed red meat (e.g. pork, beef), processed meat (e.g. sausage, hot dogs), milk and yogurt, cheese (e.g. American and cheddar cheese), fish and seafood (e.g. tuna, crab), beans (e.g. black and kidney beans), nuts (e.g. almonds, peanuts), and eggs. Mixed dishes were disaggregated and assigned to the appropriate protein group (e.g. the chicken portion of ‘rice and chicken’ was allocated to the unprocessed white meat category). Other relevant food items, including intakes of fruit and vegetables, whole grains, omega-3 fatty acids, and total sodium intake, were included in the models to control for other potential dietary factors associated with both protein-rich food groups and MetS outcomes.^37–40

MetS Components

The six MetS components from the National Cholesterol Education Program/Adult Treatment Panel III definition included fasting glucose (mg/dL; to convert to mmol/L, multiply by 0.0555), HDL cholesterol (mg/dL), triglycerides (mg/dL), systolic blood pressure (mmHg), diastolic blood pressure (mmHg), and waist circumference (cm).⁴¹ Briefly, participants were classified with MetS if they had three or more of the following outcomes: 1) waist circumference ≥ 102 cm for men or ≥ 88 cm for women, 2) HDL-C < 40 mg/dL for men or < 50 mg/dL for women, 3) triglycerides ≥ 150 mg/dL or use of lipid-lowering medication, 4) fasting glucose ≥ 100 mg/dL or use of medication, and 5) systolic blood pressure ≥ 130 mm/Hg, diastolic blood pressure ≥ 85 mmHg, or use of hypertension medication. The components were measured at baseline and 2-year, and the 2-year change for each component was calculated as the year 2 value minus the value at baseline, with positive values indicating an increase and negative values indicating a decrease. At the baseline and 2-year visits, participants were instructed to fast for 12 hours prior to the blood draw, which was conducted by a phlebotomist in their home. Serum glucose was measured using an enzymatic kinetic reaction on the Olympus AU400e with Olympus glucose reagents (Olympus America). Blood lipids were measured from EDTA plasma with an enzymatic endpoint reaction on the Olympus AU400e with Olympus HDL-C and TG reagents (Olympus America). Systolic and diastolic blood pressures were measured in duplicate at three time points during the interview. The average of the second and third readings were used as the recorded value. Waist circumference was measured in duplicate,⁴² and the average of the two values was recorded.

Covariates

Baseline covariates considered were age, sex, medications (i.e. current use of medication prescribed for diabetes, hypertension, and hyperlipidemia at baseline), outcome measure at baseline, education, poverty income ratio, psychological acculturation, smoking status, alcohol use, physical activity, servings of fruits and vegetables, servings of whole grains, intake of omega-3 fatty acids, and total sodium intake. Age, sex, medications, education, household income, alcohol use, and smoking status were self-reported in the questionnaire by the participants. Smoking was evaluated on a never, former, vs. current scale, and alcohol was assessed as the average amount of alcohol consumed over the past year (none, 0 drinks per day; moderate, ≤ 1 drink per day for women or 2 drinks per day for men; or heavy, more than 1 or 2 drinks per day for women and men, respectively). Psychological acculturation was measured using a continuous score that determined attachment to Hispanic and mainland US culture, where a higher score implied more attachment to the US.⁴³ Physical activity was measured as a continuous score adapted from the Paffenbarger questionnaire in the Harvard Alumni Activity Survey, where a higher score indicated more physical activity.^44,45 The aforementioned other relevant food items, calculated as daily intake from the FFQ, were considered as covariates. Because sodium intake is often not accurately assessed in FFQs,⁴⁶ total sodium was represented in quintiles as a ranking of the population’s intake.

Statistical Methods

Baseline characteristics were compared by protein intake status using Wilcoxon rank sum and chi-square tests. Those with less than the recommended daily intake of protein from Richter et al. (2019) (for females and males, respectively: 48 and 57 g/day for ages 25 to less than 51 years, 47 and 55 g/day for ages 51 to less than 65 years, and 57 and 67 g/day for 65 years and older) were compared with those who consumed greater than the recommended daily intake.⁴⁷ Baseline and 2-year characteristics were compared using Wilcoxon matched-pairs sign rank tests for the continuous measures, McNemar’s tests for the dichotomous categorical variables, and marginal homogeneity tests for the other categorical variables with more than two options. Medians (interquartile range (IQR)) were presented for continuous variables, and frequencies for categorical variables. Daily intake (median (IQR)) of each of the protein groups was adjusted for total energy intake using the residual method and compared by protein intake status using a Wilcoxon rank sum test. The percentages of daily intakes of each protein group (g/day) were calculated as percentage of total protein intake and of total energy intake and then ranked using PROC RANK in SAS version 9.4.⁴⁸

The association between each continuous protein food group and total protein intake with continuous change in each MetS outcome was assessed using linear mixed models. β coefficients and 95% confidence intervals (CI) are reported, describing the change in the outcome given a one serving increase in the protein group. Models were adjusted for total energy intake, sex, age, education, baseline outcome, smoking status, alcohol frequency, physical activity, psychological acculturation, the sum of fruit and vegetable intake, omega-3 fatty acid intake, whole grain intake, medication (for blood pressure, triglycerides, and glucose only), and total sodium intake (for blood pressure only). In sensitivity analyses, models were adjusted for all other protein groups simultaneously. In a secondary analysis, the associations between each protein food group and the odds of MetS at follow-up were assessed using logistic regression models adjusted for the same covariates described above (except medications), in 337 participants after excluding individuals with MetS at baseline (n=914). The level of statistical significance was < 0.05. All statistical analyses were conducted using SAS version 9.4 and Stata version 16.1.^48,49

Results

At baseline, 73% of the sample were female, and the median (IQR) age was 56 (51, 63) y (Table 1). The median (IQR) of total protein intake at baseline was 81.5 (58.4, 108) g/day (Table 2). Statistically significant differences between participants who consumed less than the recommended daily total protein intake (versus equal or more) at baseline were observed for age, education, alcohol frequency, physical activity score, and hypertension medication. Additionally, 2-year values for age, smoking status, alcohol frequency, physical activity score, fasting plasma glucose, diastolic blood pressure, HDL cholesterol, waist circumference, BMI, and use of medications for hypertension, hyperlipidemia, and diabetes were all significantly different from baseline values (Table 1). After energy adjustment, participants who consumed less than the recommended daily total protein intake (versus equal or more) had significantly higher intake at baseline of processed meat, eggs, nuts, and total carbohydrates, but lower intake of unprocessed poultry, total fat, total energy, vegetables, omega-3 fatty acids, and total protein (Table 2). Unprocessed red meat, unprocessed poultry, and milk and yogurt were the top three protein groups contributing to overall protein intake, while cheese, eggs, and nuts were the bottom three (Table 3). Milk and yogurt, unprocessed red meat, and beans were the top contributors to total energy intake, while fish and seafood, eggs, and nuts contributed the least. Examples of food items included in each protein group are also described in Table 3. The nine protein groups analyzed captured 66% of total protein intake and 31% of total energy intake.

Table 1:

Demographic characteristics of 1,126 Puerto Rican adults residing in Boston, Massachusetts from the Boston Puerto Rican Health Study

	Total; Baseline	Less than recommended protein intakea at baseline	Greater than or equal to recommended protein intake at baseline	Total; 2 year follow-up
Total, n (%)	1126 (100)	184 (16.3)	942 (83.7)	-
Sex, n (%)
Female	821 (72.9)	135 (73.4)	686 (72.8)	-
Male	305 (27.1)	49 (26.6)	256 (27.2)	-
Age, median (IQR) b	56.0 (51.0, 63.0)	62.0 (55.0, 68.0)	56.0 (51.0, 62.0)*	58.5 (53.0, 65.0)**
Education, n (%)
Less than high school	733 (65.1)	137 (74.5)	596 (63.3)*	-
High school diploma or equivalent	226 (20.1)	29 (15.8)	197 (20.9)
Some college	167 (14.8)	18 (9.8)	149 (15.8)	-
Income poverty ratio, median (IQR)	92.3 (73.6, 132)	89.5 (75.2, 116)	93.4 (73.1, 135)	96.1 (77.2, 120)
Smoker, n (%)
Never	518 (46.0)	87 (47.3)	431 (45.8)	514 (45.7)**
Former	340 (30.2)	52 (28.3)	288 (30.6)	369 (32.8)
Current	268 (23.8)	45 (24.5)	223 (23.7)	243 (21.6)
Alcohol Frequency, n (%)
None	628 (55.8)	131 (71.2)	497 (52.8)*	749 (66.6)**
Moderate	423 (37.6)	47 (25.5)	376 (39.9)	325 (28.9)
Heavy	75 (6.7)	6 (3.3)	69 (7.3)	50 (4.5)
Physical activity score c , median (IQR)	30.4 (28.2, 33.2)	29.4 (27.7, 31.8)	30.6 (28.4, 33.6)*	30.6 (28.3, 34.3)**
Psychological acculturation score d , median (IQR)	18.0 (12.0, 23.0)	18.0 (11.0, 23.0)	18.0 (12.0, 23.0)	18.0 (10.0, 22.0)
Hypertension Medication, n (%)
Current user	626 (55.8)	117 (63.6)	509 (54.3)*	685 (61.0)**
Hyperlipidemia Medication, n (%)
Current user	479 (42.7)	90 (48.9)	389 (41.5)	554 (49.3)**
Diabetes Medication, n (%)
Current user	361 (32.2)	59 (32.1)	302 (32.2)	410 (36.5)**
Fasting plasma glucose (mg/dL) e , median (IQR)	103 (92.0, 128)	102 (91.0, 128)	103 (92.0, 128)	102 (90.0, 124)**
Systolic BP (mmHg) f , median (IQR)	134 (121, 146)	136 (124, 148)	133 (121, 146)	134 (122, 148)
Diastolic BP (mmHg) f , median (IQR)	80.5 (73.5, 87.8)	79.3 (72.5, 86.0)	80.9 (73.8, 88.0)	80.0 (72.8, 86.8)**
Triglycerides (mg/dL), median (IQR)	136 (100, 194)	138 (94.0, 187)	135 (100, 194)	139 (98.0, 188)
HDL cholesterol (mg/dL)g median (IQR)	43.0 (36.0, 51.0)	43.0 (37.0, 52.0)	43.0 (36.0, 51.0)	44.0 (38.0, 53.0)**
Waist circumference (cm), median (IQR)	101 (92.1, 111)	101 (91.4, 111)	101 (92.1, 111)	102 (93.5, 113)**
BMI (kg/m2)h, median (IQR)	31.1 (27.6, 36.1)	30.2 (26.7, 35.9)	31.3 (27.8, 36.2)	31.3 (27.1, 35.7)**

^aRecommended daily protein intake from Richter et al. (2019) for females is 48 g/day for ages 25 to less than 51 years, 47 g/day for ages 51 to less than 65 years, and 57 g/day for ages. 65 years and older. Recommended daily protein intake for males is 57 g/day for ages 25 to less than 51 years, 55 g/day for ages 51 to less than 65 years, and 67 g/day for ages. 65 years and older.

^bIQR=inter-quartile range

^cPhysical activity score is the weighted sum of the scales for sleeping, laying down, vigorous, moderate, light, and sitting activity

^dPsychological acculturation score is the sum of ten questions surrounding one’s attachment to mainland Puerto Rican culture vs. US culture. The score ranges from 0 to 50, with 0 meaning identifying with Puerto Rican culture and 50 meaning identifying with US culture.

^eTo convert mg/dL to mmol/L, multiply by 0.0555

^fBP=blood pressure

^gHDL=high-density lipoprotein

^hBMI=body mass index

^*Significantly different at p<0.05 between categories of g/day of total protein intake at baseline

^**Significantly different at p<0.05 from baseline value for all participants

Table 2:

Baseline dietary intake of protein sources and other selected food groups and nutrients among Puerto Rican adults residing in Boston, Massachusetts^a

	Total	Less than recommended protein intakeb at baseline	Greater than or equal to recommended protein intake at baseline
Total; n (%)	1126 (100)	184 (16.3)	942 (83.7)
	Median (IQR)c	Median (IQR)c	Median (IQR)c	p-valued
Unprocessed red meat (servings/day)	0.25 (0.16, 0.39)	0.24 (0.19, 0.32)	0.25 (0.14, 0.42)	0.76
Unprocessed poultry (servings/day)	0.20 (0.12, 0.35)	0.16 (0.12, 0.23)	0.22 (0.11, 0.37)	<0.001
Processed meat (servings/day) %	0.37 (0.20, 0.59)	0.41 (0.32, 0.49)	0.36 (0.17, 0.62)	0.03
Milk and yogurt (servings/day)	1.25 (0.69, 1.94)	1.22 (0.82, 1.76)	1.26 (0.64, 2.00)	0.95
Cheese (servings/day)	0.42 (0.24, 0.67)	0.42 (0.32, 0.55)	0.41 (0.21, 0.70)	0.24
Eggs (servings/day)	0.13 (0.07, 0.25)	0.13 (0.10, 0.23)	0.12 (0.05, 0.26)	0.01
Fish and seafood (servings/day)	0.14 (0.08, 0.23)	0.14 (0.11, 0.19)	0.14 (0.07, 0.24)	0.49
Beans (servings/day)	0.30 (0.17, 0.47)	0.32 (0.23, 0.42)	0.29 (0.15, 0.50)	0.08
Nuts (servings/day)	0.08 (0.01, 0.14)	0.11 (0.09, 0.14)	0.06 (0.00, 0.14)	<0.001
Total carbohydrates (g/day)	262 (241, 286)	270 (255, 286)	260 (236, 286)	<0.001
Total fat (g/day)	74.7 (67.2, 81.7)	72.4 (67.4, 77.5)	75.5 (67.1, 82.8)	0.001
Energy (kcal/day)	1933 (1406, 2530)	1059 (901, 1288)	2124 (1639, 2687)	<0.001
Whole grains (servings/day)	0.72 (0.30, 1.38)	0.62 (0.32, 1.06)	0.75 (0.28, 1.44)	0.09
Vegetables (servings/day)	1.62 (1.09, 2.29)	1.43 (1.05, 1.81)	1.67 (1.10, 2.34)	0.001
Fruit (servings/day)	0.40 (0.24, 0.81)	0.41 (0.32, 0.65)	0.39 (0.24, 0.86)	0.45
Omega-3 fatty acids (g/day)	1.57 (1.37, 1.84)	1.48 (1.37, 1.66)	1.58 (1.36, 1.87)	0.001
Sodium (g/day)	4469 (3923, 5096)	4428 (4119, 4902)	4476 (3856, 5165)	0.86
Total protein (g/day)	81.5 (58.4, 108)	42.1 (35.3, 47.8)	87.9 (69.8, 116)	<0.001

^aAll values aside from energy and total protein are adjusted for total energy intake. All protein groups (servings/day) and other dietary measures were assessed at baseline and compared by the recommended (g/day) value of protein. Intake of food groups and nutrients were assessed using semi-quantitative food frequency questionnaire adapted to the Puerto Rican dietary pattern that estimated dietary intake over the past 12 months.

^bRecommended daily protein intake from Richter et al. (2019) for females is 48 g/day for ages 25 to less than 51 years, 47 g/day for ages 51 to less than 65 years, and 57 g/day for ages. 65 years and older. Recommended daily protein intake for males is 57 g/day for ages 25 to less than 51 years, 55 g/day for ages 51 to less than 65 years, and 67 g/day for ages. 65 years and older.

^cIQR=inter-quartile range

^dp-values calculated from the Wilcoxon rank sum test

Table 3.

Ranking of food groups based on contribution to total protein and total energy intake in Puerto Rican adults residing in Boston, Massachusetts^a

	Mean intake (g/day)	Examples of food items in the food-frequency questionnaire	% contribution to total protein intake	% contribution to total energy intake
Unprocessed red meat	1.41	Pork, hamburger, beef steak, beef brisket, beef cubes	13.3	4.78
Unprocessed poultry	1.96	Chicken, turkey	13.0	3.90
Milk and yogurt	10.7	Milk (1%, 2%, skim, whole), cream, chocolate milk, yogurt, Ultra Slim-Fast	9.20	6.03
Fish and seafood	1.2	Crab, clams, crayfish, haddock, lobster, salmon, sardines, scallops, shrimp, tuna, catfish, oyster, bacalao	7.93	1.97
Processed meat	1.53	Beef jerky, hot dogs, sausage, bacon, ham, salami, turkey deli style, bologna	6.58	3.11
Beans	4.61	Black beans, kidney beans, lima beans, pinto beans, white beans, kidney beans, refried beans, pink beans, cowpeas, pigeon peas, chili beans	6.12	4.26
Cheese	1.09	Cottage cheese, feta cheese, cheddar cheese, American cheese, swiss cheese, Brie cheese, cream cheese, provolone cheese, mozzarella cheese, ricotta cheese, parmesan cheese,	5.86	3.93
Eggs	1.89	Fried eggs, boiled eggs, omelet, scrambled eggs, whole eggs	2.92	1.85
Nuts	0.42	Mixed nuts, peanuts, almonds, cashews, pecans, pistachios, nuts and seeds, walnuts, peanut butter	0.91	0.99

^aProtein groups (g/day) were ranked based on their percentage contribution to total protein intake (g/day). Percentage contribution to total energy intake (kcal/day) was also calculated using the kcal/day amount for each protein group.

Adjusted associations between intake of protein-rich foods at baseline and 2-year change in MetS components were tested (Table 4). A one serving increase in processed meats was associated with a 1.28 cm increase in waist circumference (β=1.28; SE=0.63), whereas a one serving increase in fish and seafood was associated with a decrease in triglycerides of 3.47 mg/dL (β=−3.47; SE=1.39). Higher intake of unprocessed poultry was associated with a decrease in triglycerides (β=−24.5; SE=9.13). No other protein sources were significantly associated with two-year changes in any of the six MetS components. Total protein intake at baseline was not associated with changes in MetS components (Table 5).

Table 4.

Associations between dietary protein food groups and 2-year changes in metabolic syndrome components among Puerto Rican adults residing in Boston, Massachusetts^a

β (95% CI) b	Unprocessed poultry (servings/day)	Unprocessed red meat (servings/day)	Processed meat (servings/day)	Milk and yogurt (servings/day)	Cheese (servings/day)	Fish and seafood (servings/day)	Beans (servings/day)	Nuts (servings/day)	Eggs (servings/day)
Fasting plasma glucose (mg/dL)	2.64 (−7.37, 12.7)	−5.91 (−15.3, 3.45)	3.98 (−2.13, 10.1)	0.91 (−1.27, 3.09)	−0.45 (−5.51, 4.62)	5.01 (−8.45, 18.5)	−3.07 (−11.3, 5.12)	4.18 (−2.95, 11.3)	3.76 (−7.70, 15.2)
Triglycerides (mg/dL)	−24.5 (−42.5, – 6.63)*	−5.48 (−22.3, 11.4)	−0.07 (−10.9, 10.7)	2.65 (−1.19, 6.48)	−0.40 (−9.62, 8.82)	−10.9 (−35.5, 13.7)	5.58 (−9.21, 20.4)	1.22 (−11.6, 14.0)	−17.6 (−38.1, 2.86)
Waist circumference (cm)	0.57 (−1.47, 2.60)	1.65 (−0.24, 3.54)	1.28 (0.04, 2.52)*	−0.04 (−0.47, 0.39)	0.55 (−0.46, 1.55)	−3.47 (−6.20, – 0.75)*	−0.37 (−2.01, 1.26)	0.07 (−1.38, 1.52)	−0.03 (−2.42, 2.36)
Systolic BPc	2.02	−0.74	−1.06	0.15	−1.09	0.95	1.39	−1.34	2.26
(mmHg)	(−1.93, 5.97)	(−4.52, 3.05)	(−3.52, 1.39)	(−0.70, 1.00)	(−3.08, 0.90)	(−4.43, 6.33)	(−1.81, 4.59)	(−4.12, 1.45)	(−2.25, 6.77)
Diastolic BPc (mmHg)	−0.91 (−3.12, 1.30)	−0.31 (−2.40, 1.79)	−0.75 (−2.11, 0.62)	0.22 (−0.25, 0.69)	−0.68 (−1.80, 0.43)	2.10 (−0.90, 5.11)	−0.00 (−1.80, 1.79)	−0.56 (−2.10, 0.99)	1.09 (−1.43, 3.60)
HDL cholesterold (mg/dL)	0.78 (−1.10, 2.66)	1.11 (−0.69, 2.92)	0.54 (−0.61, 1.69)	0.11 (−0.30, 0.52)	0.01 (−0.96, 0.98)	1.74 (−0.83, 4.30)	−0.56 (−2.12, 0.99)	−0.37 (−1.77, 1.02)	1.17 (−1.00, 3.35)

^*p < 0.05

^aAssessed using linear mixed models adjusted for total energy intake, sex, age, education, baseline outcome, smoking, alcohol intake, physical activity, psychological acculturation, fruit and vegetable intake score, omega-3 fatty acid intake, whole grain intake, medication (for blood pressure, triglycerides, and glucose only), and sodium intake (for blood pressure only). Reported values are the estimated β (95% confidence interval).

^bCI=confidence interval

^cBP=blood pressure

^dHDL=high-density lipoprotein

After adjusting for all other protein groups, the association between processed meat and waist circumference was attenuated and no longer significant (Table 6). The association between unprocessed poultry and decreased triglycerides, as well as fish and seafood and decreased waist circumference remained significant. No protein food group was significantly associated with the odds of MetS at follow-up (Table 7).

Discussion

The aim of this study was to assess associations between intake of various dietary protein sources and two-year changes in MetS components in Puerto Ricans living in the mainland US. Higher intake of fish and seafood was associated with a decrease in waist circumference, whereas higher intake of processed meat was associated with an increase in waist circumference over two years. Unprocessed poultry was associated with a decrease in triglycerides. To our knowledge, this is the first longitudinal study that demonstrates longitudinal associations between dietary protein sources and cardiometabolic risk factors in US Hispanics.

The observed association between higher intake of processed meat and increased waist circumference is in agreement with several previous studies.^{14,25,26,50,51} A meta-analysis conducted by Rouhani et al. found processed meat to be associated with higher risk of elevated waist circumference as well as obesity.¹⁴ Among Hispanics in the Adventist Multiethnic Nutrition Study, non-vegetarians were found to have significantly higher waist circumference than vegetarians.⁵² A study of Danish adults found an association between processed meat and increased waist circumference in women, but not men.⁵⁰ Of note, adjustment for all protein groups attenuated the observed association for processed meat, suggesting that other protein sources may confound this association; however, simultaneously adding correlated foods may over-adjust the models.

The aforementioned study of Danish adults did not find the association between fish and decreased waist circumference that was observed in the present study.⁵⁰ Discrepancies in results may be due to overall differences in dietary patterns and specific types of fish and seafood consumed by the two cohorts. Some studies have analyzed fatty and lean fish separately in relation to MetS components. One study found fatty fish to be associated with an increase in waist circumference over 13-years, with no significant association between lean fish and change in waist circumference after adjustment for potential confounders.⁵¹ Lastly, one cross-sectional study in middle-aged Norwegians found significantly lower waist circumference for individuals consuming the highest quartiles of both total and lean fish when compared with the lowest quartiles.²⁶ Median intake of fish and seafood in this cohort was low (0.14 servings/day) and thus, lean and fatty fish could not be analyzed separately. It is possible that the association may vary when stratifying this protein source. Of note, the recommended amount of fish consumption per day is 0.25 g⁵³, thus, strategies to help Puerto Rican adults reach the recommended fish intake may translate into a healthier waist circumference.

There are several possible mechanisms that may explain the associations between processed meat and fish/seafood with waist circumference. A study by Nettleton et al. found that dietary pattern was an effect modifier in the association between several single nucleotide polymorphisms and BMI-adjusted waist-hip ratio, and a healthier diet (containing lower intake of processed meats and higher intake of fish) strengthened this association.⁵⁴ Additionally, processed meats are highly energy-dense, which has been associated with obesity.^55,56 Fish has higher levels of n-3 fatty acids, which have been linked to lower rates of obesity.^57,58 Further investigation into the mechanisms underlying these associations could be beneficial to develop targeted interventions and dietary plans.

Unprocessed poultry was significantly associated with a 2-year decrease in triglycerides. This association has not been noted widely in previous studies. One randomized controlled trial by Davidson et al. randomized participants to lean red meat (beef, veal, pork) or lean white meat (fish and poultry), but did not find a difference in triglycerides at baseline or follow-up (36 weeks).⁵⁹ However, this trial did not evaluate the isolated effect of poultry on blood lipids. A meta-analysis found that red meat intake was associated with lesser decreases in triglycerides than comparison diets, many of which included poultry.⁶⁰ While the observed association between poultry and changes in triglycerides has not been widely noted, there are documented health benefits of consuming poultry as part of a healthy diet. Poultry contains a number of vitamins and minerals, such as thiamin, vitamin B6, iron, zinc, and copper, and offers easily digestible protein.⁶¹ Consumption of poultry within a broader healthy diet has been found to reduce the risk of obesity and T2D.⁶¹ Additionally, poultry is high in monounsaturated fats, which has been shown to lower triglycerides.^61,62 While this analysis was adjusted for omega-3 fatty acids, the association may have been induced by monounsaturated fatty acids in poultry. Given that the association with decreased triglycerides remained significant after adjusting for all other protein groups, it is likely that the estimate is reflective of an association with poultry rather than consumption of other proteins.

No other associations between protein groups and changes in MetS components were observed in this population. The dietary patterns of these US-residing Puerto Ricans likely differ from the populations of previous studies that have found significant associations. The amount of protein recommended for adults older than 50 y is 55–67 g/day for males and 47–57 g/day for females.⁴⁷ The median (IQR) amount of protein consumed in this population was 81 (58, 108) g/day, which is higher than recommended values. The heterogeneity of the results of this study and previous studies suggests the need for more studies of dietary protein and MetS in US Hispanic populations. Of note, the null associations of other dietary protein sources with MetS risk factors suggest that some of these protein-rich foods, such as beans, nuts, and eggs, that may be associated with reductions in risk of MetS,^63–66 may still be consumed by Puerto Ricans without worsening cardiometabolic health, especially as these sources slightly contributed to total protein intake in this population. It is important to consider that lower consumption of such protein sources may have contributed to the null findings.

This analysis has some limitations. First, while the results may be generalizable to people of Puerto Rican heritage and other ethnic groups with similar protein intake profile, there are sociocultural and environmental factors related to both protein sources and cardiometabolic factors that may vary for a same heritage group across locations (e.g., local vs. imported protein source; culinary practices), and thus it would be important to replicate these observations in other cohorts. However, this group is disproportionally impacted by MetS.^3,4 Thus, these results still hold practical significance. Also, self-reported dietary data inherently holds some bias. Nonetheless, the FFQ used in this study has been validated, is culturally relevant/appropriate, and has been used successfully in many analyses.^33–35,67 Dietary protein intake at baseline was used in this analysis, and dietary behaviors could have changed over the two-year follow-up period. However, it has previously been shown that diet and lifestyle behaviors were relatively stable in this study population over the two years of follow-up; the mean dietary intakes were almost identical at baseline and at two years.⁶⁸ Finally, the large number of individuals with MetS at baseline dampened the ability to assess incident MetS at the 2-year follow-up.

This study also has a number of strengths. First, this cohort contains a large sample size of minority participants who have been underrepresented in epidemiological studies. The study collected data on numerous behaviors and health outcomes, allowing adjustment for many potential confounders, including other dietary sources. The FFQ and food grouping in this study captured 66% of the total protein intake; other possible protein sources typically classified in other dietary food groups (such as oatmeal, corn, and other grains) should be analyzed in future studies. Additionally, the prospective approach used is beneficial for enhancing likelihood of causal relationships with MetS outcomes, by reducing the potential for reverse causality. Of note, the current findings support continued strategies for healthy protein intake in order to achieve clinically relevant metabolic improvements. For example, this study found a 1.3 cm increase in waist circumference per serving of processed meat. Previously, a 1 cm increase was shown to be associated with a higher risk of developing new-onset type 2 diabetes, impaired fasting glucose, hypertension, and left ventricular hypertrophy.⁶⁹ Additionally, a trial-level meta-regression analysis showed a 0.84 (95% CI, 0.75, 0.94) lower risk of major vascular events per 1 mmol/L (~88.5 mg/dL) reduction in triglycerides.⁷⁰ Consumption of fish and seafood and unprocessed poultry was associated with 3.57 to 24.5 mg/dL lower triglycerides in the current study, which cumulatively could translate to long term cardiovascular benefits.

Conclusions

This study of Puerto Ricans living in the mainland US found statistically significant and clinically-relevant associations between higher intake of fish and seafood and lower intake of processed meat and 2-year decreases in waist circumference, and higher intake of unprocessed poultry and 2-year decreases in triglyceride concentration. Additional studies are needed to confirm these findings with other health outcomes influenced by protein intake and other specific types of proteins not assessed in this study.⁷¹ If confirmed by future analyses, these results suggest that culturally tailored diet interventions should target lowering intake of processed meats and increasing intake of fish and unprocessed poultry to improve cardiometabolic health in Puerto Rican adults.

Research snapshot.

Research question:

Is intake of various sources of protein-rich foods associated with the six metabolic syndrome components over two years in Puerto Ricans residing in the US mainland?

Key findings:

Using longitudinal data from 1,126 participants in the Boston Puerto Rican Health Study, higher intake of fish and seafood was found to be significantly associated with decreased waist circumference over two years, whereas higher intake of processed meat was associated with increased waist circumference. Higher intake of unprocessed poultry was significantly associated with decreased triglycerides over two years. Associations with other dietary protein sources were not significant. Cardiometabolic health may be improved by consuming less processed meats and more fish and seafood as well as unprocessed poultry.

Abbreviations:

MetS

Metabolic Syndrome

T2D

type 2 diabetes

US

United States

AHEI

Alternate Healthy Eating Index

FFQ

food frequency questionnaire

IQR

interquartile range