Assessment of Dominant Dietary Patterns in Hemodialysis Patients Using Factor Analysis Method and Their Relationship With Anthropometric Indices and Metabolic Parameters: An Analytical Cross-Sectional Study

Abstract

Chronic kidney disease (CKD) often progresses to end-stage renal disease (ESRD), requiring interventions like dialysis. Nutrition plays a key role in CKD management by affecting metabolic and anthropometric outcomes. This study assessed dominant dietary patterns in hemodialysis patients using factor analysis and their relationships with anthropometric and metabolic parameters. This cross-sectional study included 300 adult hemodialysis patients selected through systematic sampling from dialysis centers in Ahvaz, Iran. Dietary intake was evaluated using a validated 168-item Food Frequency Questionnaire (FFQ). Principal Component Analysis (PCA) identified 3 dietary patterns: Western, Mixed, and Traditional. Anthropometric indices (BMI, waist circumference) and metabolic parameters (lipid profiles, renal markers) were assessed. Multiple linear regression adjusted for confounders was applied. The Western Pattern was positively associated with higher BMI (+1.41 kg/m², P < .001), WC (+3.67 cm, P < .001), total cholesterol (+11.81 mg/dL, P < .001), and LDL-C (+9.54 mg/dL, P < .001). The Mixed Pattern showed protective effects, reducing WC (−3.30 cm, P < .001), TC (−6.13 mg/dL, P = .004), LDL-C (−5.01 mg/dL, P = .001), and uric acid (−0.75 mg/dL, P = .03). The Traditional Pattern was linked to lower BMI (−1.57 kg/m², P < .001), WC (−3.91 cm, P < .001), improved HDL-C (+1.73 mg/dL, P = .02), and reduced systolic blood pressure (−3.57 mmHg, P < .001). Dietary patterns significantly impact metabolic and anthropometric health in hemodialysis patients. Adopting culturally relevant dietary patterns like the Traditional Pattern may improve outcomes. Further longitudinal studies are warranted.

Graphical Abstract

Introduction

Chronic kidney disease (CKD) is a major global public health problem with a steadily rising incidence and prevalence worldwide. The progressive loss of kidney function frequently leads to end-stage renal disease (ESRD), requiring renal replacement therapy (RRT) such as hemodialysis or kidney transplantation.^1,2 Globally, CKD affects approximately 10% to 15% of adults, with lifetime risk estimates approaching 50% in some high-income countries.^2,3 In Iran, prevalence rates range from 6.5% to 23.7%, depending on screening methods, age, and risk factor distribution. ⁴ The number of patients on RRT continues to increase rapidly, particularly in Asia, where it is projected to exceed 4.5 million by 2030.^2,3

Nutrition is a cornerstone of CKD management and plays a critical role in slowing disease progression, reducing uremic toxin accumulation, and preventing or ameliorating complications such as hypertension, dyslipidemia, protein-energy wasting, cardiovascular disease, and mineral-bone disorders.^5,6 Although single-nutrient or single-food approaches have traditionally guided renal dietary recommendations, they fail to capture the complex synergistic and antagonistic interactions among foods and nutrients consumed in combination. ⁷ In contrast, the analysis of overall dietary patterns provides a more holistic, realistic, and clinically applicable framework for understanding diet–disease relationships and for developing effective nutritional interventions.^8,9

Dietary patterns can be derived using a priori (hypothesis-driven) methods (eg, Mediterranean or DASH scores) or a posteriori (data-driven) methods such as principal component analysis (PCA) and cluster analysis. PCA is particularly valuable in culturally diverse populations because it identifies patterns specific to the studied cohort without preconceived assumptions. ⁹ Growing evidence links unhealthy dietary patterns typically high in processed and red meats, refined grains, sweets, and sodium with adverse metabolic outcomes in CKD patients, including elevated triglycerides, uric acid, inflammation, and central obesity.^{10 -12} Conversely, prudent or plant-based patterns are associated with lower CKD incidence, slower decline in estimated glomerular filtration rate (eGFR), and improved cardiometabolic profiles.^11,13,14 However, most studies have been conducted in Western or East Asian populations, and data on dietary patterns among maintenance hemodialysis patients, particularly in Middle Eastern countries, remain scarce. ¹⁵

To address this gap, the present analytical cross-sectional study was conducted among Iranian adults undergoing maintenance hemodialysis with the following specific objectives:

1. To identify dominant dietary patterns using principal component analysis (PCA);

2. To evaluate the associations of these patterns with anthropometric indices (body mass index, waist circumference) and key metabolic parameters (lipid profile, glycemic status, blood pressure, renal and hepatic markers);

3. To provide culturally relevant evidence to support the development of tailored nutritional interventions for this high-risk population.

Methods

Participants and Study Design

In this cross-sectional study, adult hemodialysis patients from 4 dialysis centers in Ahvaz, Iran, were assessed between March and September 2023. A comprehensive list of all hemodialysis centers in Ahvaz was initially obtained from Ahvaz Jundishapur University of Medical Sciences. The researchers then visited these 4 centers to compile a list of all hemodialysis patients. From this pool, 452 patients who met the inclusion criteria were identified. The centers were arranged alphabetically, and the patients’ names from each center were listed accordingly. Using systematic sampling, 300 patients were selected from the 452 eligible individuals. The flow diagram of the sampling process is illustrated in Figure 1. This systematic random sampling strategy was deliberately chosen to ensure proportional representation across centers, reduce selection bias, and increase the external validity of the findings to the hemodialysis population of southwest Iran.

Figure 1.

Flow diagram of participant selection and inclusion in the study. Systematic random sampling from 732 registered hemodialysis patients at 4 dialysis centers in Ahvaz, Iran, resulting in a final analytical sample of 300 participants after application of strict inclusion/exclusion criteria.

The inclusion criteria required participants to be adults (aged ⩾ 18 years), undergoing dialysis for at least 1 year, receiving dialysis 3 times a week, and willing and able to participate in the study. Exclusion criteria included the presence of underlying conditions such as thyroid or liver disease, cancer, HIV, infectious or inflammatory diseases, as well as smoking, alcohol or hookah use, recent use of steroid or non-steroidal anti-inflammatory drugs, antioxidant supplementation within the past 3 months, participation in weight change programs, incomplete responses to more than 70 items of the Food Frequency Questionnaire (FFQ), and total energy intake below 800 kcal or above 4200 kcal. These rigorous exclusion criteria were applied to minimize residual confounding from acute inflammation, catabolic states, pharmacological interference with lipid and glucose metabolism, and implausible dietary reporting, which are common limitations in nutritional epidemiology studies among dialysis patients.

All participants were on bicarbonate hemodialysis using polysulfone capillary dialyzers, with dialysis sessions conducted 3 times a week for 4 hours each. Written informed consent was obtained from all participants. The study received approval from the ethics committee of Ahvaz Jundishapur University of Medical Sciences (IR.AJUMS.REC.1401.483) and adhered to the principles of the Declaration of Helsinki. Uniformity of dialysis prescription across participants was essential to reduce variability in fluid status, uremic toxin burden, and post-dialysis weight, all of which could otherwise confound anthropometric and biochemical outcomes.

Assessment of Dietary Intake

Data on dietary intake were gathered using a semi-quantitative Food Frequency Questionnaire (FFQ) comprising 168 items. ¹⁶ Participants were asked to indicate the frequency of consumption for each food item over the past year. Depending on the food type, frequencies were reported as daily, weekly, monthly, or yearly consumption. Standard portion sizes, as well as measurements reported using household scales, were converted into grams using the Nutritionist 4 Home Scales Guide. Subsequently, the daily intake of each food item was calculated and expressed in grams per day. The FFQ was administered through standardized, face-to-face interviews conducted by a single trained nutritionist to enhance accuracy, reduce social desirability bias, and accommodate varying literacy levels (Table 1).

Table 1.

Food Groups Used in the Factor Analysis.

Food groups	Food items
Refined grains	Lavash bread, baguette, toast, rice, pasta, vermicelli, noodles, wheat flour, biscuits, crackers, Yazdi cake, homemade cake, other cakes.
Whole grains	Barbari bread, sangak bread, taftoon bread, corn and corn on the cob, barley, bulgur.
Legumes	Lentils, beans, chickpeas, broad beans, soybeans, mung beans, split peas.
Red meat	Beef, lamb, minced meat, stew.
Chicken meat	Chicken with skin, skinless chicken.
Fish meat	Fish, canned tuna.
Fast foods and processed meats	Sausage, pizza, french fries, pastries, hamburger, cold cuts.
Eggs	Egg
Organ meats	Heart, liver, and kidneys, tripe, tongue, brain, head, feet.
Low-fat dairy products	Skim milk, low-fat milk, regular yogurt, cheese.
High-fat dairy products	Full-fat milk, cocoa milk, chocolate milk, condensed yogurt, full-fat yogurt, creamy yogurt, cream cheese, cream and skimmed milk, traditional ice cream, non-traditional ice cream, doogh, whey curd.
Fats and hydrogenated oils	Hydrogenated vegetable oil, butter, caramel cream, mayonnaise, suet, margarine, animal oil.
Vegetable oils (excluding olive oil)	Liquid oil.
Olives and olive oil	Green olives, olive oil.
Nuts	Peanuts, almonds, walnuts, pistachios, hazelnuts, sunflower seeds.
Vegetables	Lettuce, cucumber, cooked green beans, cabbage, bell peppers, fresh herbs, zucchini, cooked eggplant, cooked green peas, raw carrots, cooked carrots, raw onions, fried onions, raw spinach, cooked spinach, cooked mushrooms, cooked pumpkin, cooked celery, garlic, cooked potato, turnip, tomato, cooked vegetables.
Fruit	Sweet lime, lemon, pear, cornelian cherry, watermelon, cherry, sour cherry, apple, apple juice, peach, fresh fig, grapes, mandarin orange, strawberry, fresh pineapple, canned pineapple, raisins, fresh mulberries, dried mulberries, persimmon, dates, cantaloupe, melon, apricot, nectarine, green plums, dried figs, kiwi, grapefruit, grapefruit juice, orange, orange juice, pomegranate, plum, banana, cantaloupe juice, Garmak melon (Iranian cantaloupe), dried peach slices, dried apricot slices, canned fruits.
Snacks	Puffed snacks, chips.
Sugars	Candy, nougat, candies, sugar cubes, sugar, soft drinks, dry sweets, wet sweets, sohan, chocolate, homemade halva, tahini halva, nabat (sugar crystal), honey, jam.
Pickles and salty foods	Pickles, salty snacks, salt, pickled cucumber.
Tea and coffee	Tea, coffee.
Flavorings	Black pepper, lemon juice, tomato sauce.

To estimate energy, macronutrient, and micronutrient intake, the gram equivalents of food items reported in the FFQ were analyzed using a food composition table specifically tailored to Iranian dietary patterns. This approach ensured accurate determination of nutrient intake based on the participants’ reported food consumption. Use of a locally adapted Iranian food composition database was critical because many traditional dishes and cooking oils commonly consumed in the region are not accurately represented in international databases.

Identification and Categorization of Dietary Patterns

Dietary patterns represent habitual combinations of foods consumed by individuals, encompassing the type, quantity, variety, and proportion of foods and beverages consumed over time. Using data from the Food Frequency Questionnaire (FFQ) and Principal Component Analysis (PCA), 3 dominant dietary patterns were identified in the study population:

1. Western Dietary Pattern: This pattern is characterized by a high intake of processed food items. Foods predominantly contributing to this pattern include poultry, whole grains, flavor enhancers, fish, fast food, processed meats, hydrogenated fats and oils, low-fat dairy products, olives and olive oil, and non-olive vegetable oils. The emphasis on processed food consumption reflects modern dietary habits influenced by urbanization and globalization.

2. Mixed Dietary Pattern: The mixed dietary pattern incorporates a variety of food groups, reflecting a balanced but diverse diet. Major contributors to this pattern include salty snacks, fruits, nuts, organ meats, simple sugars, sweets, and high-fat dairy products. This pattern highlights a combination of healthy and less healthy food choices, showcasing transitional dietary behaviors influenced by personal preferences or cultural practices.

3. Traditional Dietary Pattern: The traditional dietary pattern reflects food choices rooted in local cultural practices and traditional cooking methods. Key contributors to this pattern include high-fat dairy products, vegetables, eggs, red meat, tea and coffee, legumes, and refined grains. This pattern represents a reliance on staple foods commonly consumed in the region, with limited inclusion of processed or modern food items.

Each dietary pattern was defined by the food groups with the highest positive loadings in the PCA, which were identified based on their contribution to the total variance explained by the model. These patterns served as independent variables for evaluating their associations with metabolic and anthropometric indices in subsequent statistical analyses. The 3 patterns align well with the ongoing nutrition transition observed in Middle Eastern countries: increasing penetration of Western processed foods, coexistence of healthy and unhealthy items (Mixed), and persistence of culturally rooted eating habits (Traditional).

Principal Component Analysis (PCA) Method

Principal Component Analysis (PCA) is a statistical technique used to reduce the dimensionality of data while preserving its variability. It identifies underlying patterns or components by examining the correlations among multiple variables—in this case, food groups derived from the Food Frequency Questionnaire (FFQ).^17,18 PCA was preferred over cluster analysis because it generates continuous factor scores suitable for linear regression and is the most widely accepted method for deriving a posteriori dietary patterns in nutritional epidemiology.

1. Data Preparation: Food items reported in the FFQ were grouped into predefined categories based on nutritional and cultural similarities. Examples include fruits, vegetables, grains, dairy, meats, oils, and processed foods. Items that did not fit standard groups or reflected unique dietary behaviors were categorized separately. Intake amounts for each food group were standardized (z-scores) to ensure comparability across participants. ¹⁹

2. Extraction of Components: PCA was applied to the standardized food group data to extract dietary patterns. Components were identified based on their ability to explain the shared variance among food groups. Each component represents a unique combination of food groups that tend to be consumed together. ¹⁷

3. Criteria for Component Selection: Components were retained if their eigenvalues were greater than 1.5, indicating that the component explains more variance than a single variable. ²⁰ A scree plot was used to confirm the number of meaningful components by identifying the point of inflection where additional components contributed minimally to variance explained. ²¹

4. Interpretation of Components: Varimax rotation, an orthogonal rotation method, was applied to maximize the interpretability of the components by simplifying the factor loadings. ¹⁸ Factor loadings indicate the strength and direction of each food group’s association with a component. Food groups with loadings greater than 0.2 were considered significant contributors to a dietary pattern. ¹⁹

5. Naming of Patterns: The identified components (dietary patterns) were named based on the food groups with the highest positive loadings, reflecting their predominant characteristics. For example, a pattern with high loadings for fish, poultry, and processed foods was termed the “Western Dietary Pattern,” while another dominated by traditional staples like legumes and tea was named the “Traditional Dietary Pattern.” ²⁰

6. Validation: The total variance explained by the retained components was calculated to assess the adequacy of the PCA model. Patterns were validated by their consistency with known dietary behaviors in the study population and prior literature.^17,21

PCA allowed the transformation of complex dietary data into a smaller set of interpretable patterns, facilitating the evaluation of their relationships with metabolic and anthropometric parameters in the target population.

The Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy (0.72) and Bartlett’s test of sphericity (P < .001) confirmed suitability for PCA. Orthogonal (varimax) rotation simplified factor structure, enhancing interpretability. Food groups with loadings ⩾ |0.3| were considered significant contributors to a pattern.^18,19 Internal consistency of patterns was assessed using Cronbach’s α (Western: α = 0.74; Mixed: α = 0.68; Traditional: α = 0.71). A stricter eigenvalue threshold (>1.5) and loading cutoff (⩾|0.3|) were applied to ensure only robust, clinically meaningful patterns were retained.

Data Collection and Measurements

At the beginning of the study, informed consent was obtained from participants, and the researcher (a nutritionist) conducted interviews to collect data. Information was gathered using a comprehensive multi-part questionnaire, which included demographic and anthropometric details (eg, age, sex, education level, marital status, occupation, race, income, smoking, alcohol consumption, medications, height, weight, body mass index [BMI], waist circumference, and hip circumference) along with metabolic parameters (eg, glycemic status, lipid profile, kidney and liver function, and blood pressure).

Dry weight was measured at the end of dialysis sessions using a body composition monitor digital scale (manufactured in Japan) with an accuracy of 0.1 kg. Participants were weighed without shoes and in light clothing. Height was recorded with a tape measure accurate to 0.5 cm, with participants standing upright against a wall. Waist circumference was measured at the upper edge of the iliac crest and below the navel, and hip circumference was recorded at the widest part of the hip using a measuring tape. BMI was calculated using the standard formula: weight (kg)/height (m²). All anthropometric measurements were taken immediately post-dialysis after achieving estimated dry weight to minimize the effect of interdialytic fluid gain and to ensure reproducibility.

Physical activity levels were assessed using the short version of the International Physical Activity Questionnaire (IPAQ). ²² This tool evaluates 3 categories of activity: vigorous (coefficient of 8), moderate (coefficient of 4), and walking (coefficient of 3.3). Total physical activity was calculated and classified into 3 levels: light (0-600 minutes per week), moderate (600-3000 minutes per week), and vigorous (over 3000 minutes per week), provided the activity lasted at least 10 continuous minutes.

Serum levels of fasting blood sugar, lipid profile, liver and kidney markers, dialysis adequacy, urine volume, 24-hour protein excretion, and serum levels of sodium, potassium, calcium, and phosphorus, as well as blood pressure, were extracted from patients’ medical records. Biochemical data were taken from routine monthly tests performed within ±2 weeks of the study visit; blood pressure values were averaged from the last 6 dialysis sessions to reduce session-to-session variability.

Dialysis adequacy, which evaluates the efficiency of toxin and waste removal during hemodialysis, was assessed due to its significant impact on patient health, quality of life, and survival. ²³ Inadequate dialysis has been associated with increased morbidity and mortality. ²⁴ Two primary measures, the urea reduction ratio (URR) and single-pool Kt/V (spKt/V), were used to determine dialysis adequacy:

1. Urea Reduction Ratio (URR):

2. URR was calculated using blood urea nitrogen (BUN) levels before and after dialysis ²⁵ : URR = [(predialysis BUN − postdialysis BUN) / predialysis BUN] × 100%

3. Single-Pool Kt/V (spKt/V): The spKt/V index measures the clearance of urea relative to its distribution volume during dialysis. It is defined mathematically as: spKt/V = −ln(1 − URR) where “ln” represents the natural logarithm.

According to the 2006 National Kidney Foundation-Kidney Disease Outcomes Quality Initiative (NKF-KDOQI) guidelines, the minimum acceptable thresholds for adequate dialysis are a spKt/V > 1.2 or URR ⩾ 65%, with target values set at 1.4 for spKt/V and 70% for URR. ²⁶

Assessment of Confounders

The body mass index (BMI) was determined by using the participants’ weight and height measurements taken at the conclusion of their dialysis session. Dialysis vintage was defined as the length of time each patient had been receiving hemodialysis (HD), measured in years. The adequacy of dialysis was evaluated through the Kt/V index, which factors in the duration of the dialysis session, post-dialysis weight, ultrafiltration volume, and serum urea levels both before and after the dialysis process. ²⁷

Potential confounders were selected a priori based on established biological and socioeconomic relevance to CKD outcomes.^5,6,23 These included: age, sex, education, marital status, physical activity (IPAQ categories), dialysis vintage (years), Kt/V (dialysis adequacy), energy intake (kcal/day), BMI, and comorbidities (diabetes, hypertension). All confounders were included in fully adjusted models (Model 3) for regression analyses. Confounders were chosen using directed acyclic graphs (DAGs) and extensive literature review to avoid over-adjustment and collider stratification bias.

Statistical Analysis

In this study, sampling was conducted systematically, and the sample size was determined based on the BMI variable from Arab et al.′s study using the following formula:

$$\mathrm{N}=\left[{\left({\mathrm{z}}_{1-\alpha /2}\right)}^2\mathrm{x}{\mathrm{sd}}^2\right]/{\mathrm{d}}^2\left(\begin{array}{l}\alpha =0.05,\mathrm{confidence}\mathrm{level}\mathrm{of}95\%,\\ \mathrm{sd}\mathrm{=}0.43,\mathrm{and}\mathrm{d}=0.05\%\end{array}\right)$$

Given an alpha level of 0.05 (confidence level of 95%), standard deviation (sdsdsd) of 0.43, and margin of error (ddd) of 0.05, the calculated sample size was 284. Accounting for a 5% attrition rate, the final sample size was set at 300 participants. ²⁸

All data were entered and analyzed using SPSS software (IBM SPSS Statistics, Armonk, USA) version 24. Quantitative data were reported as mean ± standard deviation (SD), while qualitative data were presented as frequency (percentage). For comparing qualitative variables, the chi-square test was used.

The relationships between dominant dietary patterns identified through Principal Component Analysis (PCA) and metabolic or anthropometric parameters were assessed using multiple linear regression models. These models adjusted for potential confounding factors, including age, gender, education, marital status, and physical activity. Associations were reported as beta coefficients with corresponding 95% confidence intervals. Three hierarchical models were constructed to evaluate the robustness of associations: Model 1 (crude), Model 2 (energy-adjusted), and Model 3 (fully adjusted for all listed confounders), allowing assessment of potential mediation and residual confounding.

Results

Demographic and Anthropometric Characteristics of Study Population

The study included 300 hemodialysis patients (170 men and 130 women) with a mean age of 52.96 ± 12.04 years. Detailed demographic and anthropometric characteristics are presented in Table 2.

Table 2.

Comparison of Basic Characteristics Between 2 Groups of Men and Women in Hemodialysis Patients.

Characteristics (mean (SD) or N (%))	Male (n = 170)	Female (n = 130)	Total (n = 300)	P-value
Age (years)	53.18 ± 11.95	52.67 ± 12.21	52.96 ± 12.04	.71*
Height (m)	169.65 ± 8.08	66.09 ± 12.36	166.23 ± 8.40	<.001*
Weight (kg)	73.88 ± 14.49	66.09 ± 12.36	70.51 ± 14.13	<.001*
BMI (kg/m2)	25.63 ± 4.47	25.27 ± 4.57	25.47 ± 4.51	.49*
BMI categories (N) (%)				.90**
Non-obese	139 (81.8)	107 (82.3)	246 (82)
Obese	31 (18.2)	23 (17.7)	54 (18)
WC (cm)	90.72 ± 12.31	82.26 ± 9.96	87.06 ± 12.08	<.001*
WC categories (N) (%)				.17**
Normal	139 (81.8)	98 (75.4)	237 (79)
High	31 (18.2)	32 (24.6)	63 (21)
HC (cm)	96.98 ± 8.43	101.01 ± 8.03	98.73 ± 8.48	<.001*
WHR	0.93 ± 0.04	0.81 ± 0.03	0.87 ± 0.07	<.001*
WHR categories (N) (%)				<.001**
Normal	155 (91.2)	68 (52.3)	223 (74.3)
High	15 (8.8)	62 (47.7)	77 (25.7)
WHtR	0.53 ± 0.07	0.50 ± 0.06	0.52 ± 0.07	.002*
WHtR categories (N) (%)				.26**
Normal	121 (71.2)	100 (76.9)	221 (73.7)
High	49 (28.8)	30 (23.1)	79 (26.3)
Wrist circumference (cm)	22.02 ± 2.12	17.08 ± 2.60	19.88 ± 3.38	<.001*
Neck circumference (cm)	38.61 ± 2.91	36.63 ± 4.58	37.75 ± 3.85	<.001*
Body frame size	7.77 ± 0.84	9.68 ± 1.49	8.60 ± 1.50	<.001*
Marital status (N) (%)				.61**
Single	26 (15.3)	20 (15.4)	46 (15.3)
Married	124 (72.9)	91 (70)	215 (71.7)
Widow	12 (7.1)	8 (6.2)	20 (6.7)
Divorced	8 (4.7)	11 (8.5)	19 (6.3)
Education (N) (%)				<.001**
Illiterate	81 (47.6)	88 (67.7)	169 (56.3)
Elementary	31 (18.2)	25 (19.2)	56 (18.7)
Middle-school	17 (10)	5 (3.8)	22 (7.3)
High-school	27 (15.9)	11 (8.5)	38 (12.7)
Collage	14 (8.2)	1 (0.8)	15 (5)
Race (N) (%)				.18**
Fras	38 (22.4)	24 (18.5)	62 (20.7)
Lor	65 (38.2)	41 (31.5)	106 (35.3)
Arab	67 (39.4)	65 (50)	132 (44)
Job (N) (%)				<.001**
Unemployed	44 (25.9)	20 (15.4)	64 (21.3)
Labor	85 (50)	14 (10.8)	99 (33)
Housekeeper	7 (4.1)	80 (61.5)	87 (29)
Employee	34 (20)	16 (12.3)	50 (16.7)
P.A (met-min/week)	78.50 ± 201.95	96.17 ± 231.76	86.16 ± 215.18	.48*
Duration of dialysis (year)	19.38 ± 5.42	19.9 ± 5.50	19.60 ± 5.45	.41*
FBS (mg/dL)	107.97 ± 36.17	110.18 ± 41.96	108.93 ± 38.73	.62*
TG (mg/dL)	125.66 ± 46.68	127.40 ± 47.90	126.42 ± 47.14	.75*
TC (mg/dL)	198.27 ± 37.61	196.31 ± 30.13	197.42 ± 34.53	.61*
LDL-c (mg/dL)	104.67 ± 27.69	102.40 ± 21.78	103.69 ± 25.29	.11*
HDL-c (mg/dL)	67.98 ± 11.80	69.85 ± 11.23	68.79 ± 11.58	.16*
LDL-c/HDL.c ratio	1.59 ± 0.52	1.51 ± 0.43	1.56 ± 0.48	.11*
TC/HDL ratio	2.99 ± 0.67	2.86 ± 0.56	2.93 ± 0.63	.09*
AIP	0.25 ± 0.19	0.22 ± 0.17	0.24 ± 0.18	.19*
B.SBP (mmHg)	150.46 ± 18.95	148.93 ± 19.06	149.80 ± 18.98	.49*
A.SBP (mmHg)	132.68 ± 17.12	131.59 ± 17.20	132.21 ± 17.14	.58*
B.DBP (mmHg)	92.36 ± 9.61	92.04 ± 9.74	92.22 ± 9.65	.77*
A.DBP (mmHg)	81.87 ± 8.14	81.45 ± 8.69	81.69 ± 8.37	.66*
B.PP (mmHg)	58.10 ± 10.48	56.89 ± 10.58	57.57 ± 10.53	.32*
A.PP (mmHg)	50.80 ± 9.64	50.14 ± 9.28	50.51 ± 9.48	.54*
B.MAP (mmHg)	111.73 ± 12.53	111.01 ± 12.63	111.0.41 ± 12.55	.62*
A.MAP (mmHg)	98.81 ± 11.01	98.16 ± 11.40	98.53 ± 11.16	.62*
B.BUN (mg/dL)	69.28 ± 16.66	68.45 ± 17.48	68.92 ± 17.00	.67*
A.BUN (mg/dL)	15.49 ± 3.24	15.23 ± 3.34	15.37 ± 3.28	.49*
Plasma Cr (mg/dL)	7.95 ± 1.56	7.62 ± 1.61	7.81 ± 1.58	.07*
Uric acid (mg/dL)	5.33 ± 1.60	5.20 ± 1.55	5.27 ± 1.58	.45*
eGFR	11.27 ± 2.01	9.06 ± 1.74	10.31 ± 2.19	<.001*
Dialysis adequacy	77.29 ± 2.73	77.41 ± 2.34	77.34 ± 2.56	.70*
Na (mEq/L)	141.46 ± 3.01	141.18 ± 3.04	141.34 ± 3.02	.42*
K (mEq/L)	4.81 ± 0.60	4.78 ± 0.63	4.79 ± 0.61	.68*
Ca (mg/dL)	8.33 ± 0.44	8.30 ± 0.40	8.32 ± 0.42	.62*
P (mg/dL)	4.74 ± 0.75	4.69 ± 0.78	4.71 ± 0.76	.59
AST (IU/L)	17.45 ± 8.88	17.24 ± 8.96	17.36 ± 8.90	.83*
ALT (IU/L)	20.74 ± 8.74	20.38 ± 8.98	20.58 ± 8.83	.72*
ALT/AST	1.20 ± 0.13	1.19 ± 0.15	1.20 ± 0.14	.54*
Western dietary pattern	0.09 ± 1.04	−0.12 ± 0.93	0.00 ± 1	.05*
Mixed dietary pattern	−0.02 ± 0.98	0.03 ± 1.02	0.00 ± 1	.63*
Traditional dietary pattern	0.10 ± 0.96	−0.14 ± 1.03	0.00 ± 1	.03*

Abbreviations: A, after dialysis; AIP, atherogenic index of plasma; ALP, alkaline phosphata; ALT, alanine aminotransferase; AST, aspartate aminotransferase; B, before dialysis; BMI, body mass index; BUN, blood urea nitrogen; Ca, calcium; Cr, creatinine; DBP, diastolic blood pressure; eGFR, estimated glomerular filtration rate; FBS, fasting blood sugar; HC, hip circumference; HDL-c, high-density lipoprotein cholesterol; K, potassium; LDL-c, low-density lipoprotein cholesterol; MAP, mean arterial pressure; Na, sodium; P, phosphorus; PA, physical activity; PP, pulse pressure; SBP, systolic blood pressure; TC, total cholesterol; TG, triglyceride; WC, waist circumference; WHR, waist-to-hip ratio; WHtR, waist-to-height ratio.

Data are means ± SD for quantitative variables and frequency (percent) for qualitative variables

Central obesity: WC ⩾ 88 cm for female and WC ⩾ 102 cm for male, WHtR ⩾ 0.5 for people under 40 years old and WHtR ⩾ 0.6 for people over 40 years old, and WHR ⩾ 0.8 for female and WHR ⩾ 1 for male.

General obesity: normal (BMI < 30) or obese (BMI ⩾ 30).

*

From Independent test for quantitative variables. **Chi-square for qualitative variables.

Comparison Between Men and Women

Men exhibited significantly higher values in several anthropometric measures compared to women. The mean height of male participants was 169.65 ± 8.08 cm, significantly greater than that of females (166.09 ± 8.40 cm, P < .001). Similarly, men had higher mean weight (73.88 ± 14.49 kg vs 66.09 ± 12.36 kg, P < .001), waist circumference (WC: 90.72 ± 12.31 cm vs 82.26 ± 9.96 cm, P < .001), and waist-to-hip ratio (WHR: 0.93 ± 0.04 vs 0.81 ± 0.03, P < .001). However, women showed significantly higher hip circumference (HC: 101.01 ± 8.03 cm vs 96.98 ± 8.43 cm, P < .001). No significant differences were observed in BMI categories between men and women (P = .90). Most participants (82%) were classified as non-obese, while 18% fell into the obese category.

Central and General Obesity Categories

The prevalence of central obesity based on WC was higher among women (24.6%) compared to men (18.2%), though this difference was not statistically significant (P = .17). WHR analysis revealed a significantly higher prevalence of high WHR among women (47.7%) compared to men (8.8%, P < .001). WHtR categories showed no significant differences between men and women (P = .26).

Further analysis of BMI categories indicated that while both genders had similar rates of general obesity, the distribution of central obesity markers varied substantially across age groups, with older women exhibiting the highest prevalence.

Physical Activity and Dialysis Characteristics

The mean duration of dialysis was 19.60 ± 5.45 years, with no significant differences between men and women (P = .41). Physical activity levels were generally low in both groups, with no statistically significant differences observed (P = .48).

Additional stratification by physical activity levels revealed that patients engaging in even moderate physical activity showed slightly better anthropometric profiles compared to sedentary participants, though the differences were not statistically significant.

Dominant Dietary Patterns

Principal Component Analysis (PCA) identified 3 dominant dietary patterns: Western, Mixed, and Traditional (Table 3). These patterns were derived from the factor loadings of various food groups.

Table 3.

Factor Loadings of Food Groups in Major Dietary Patterns. ^a .

Food groups	Major dietary patterns
	Western	Mixed	Traditional
Chicken	−0.671		0.291
Whole grains	0.647	0.227
Flavoring	0.639
Fish	0.588		0.222
Fast food and processed meat	0.566	0.326
Hydrogenated fats and oils group	−0.477	0.381
Low fat dairy products	−0.470	−0.278	0.293
Olives and olive oil	0.420
Vegetable oil group except olive oil	0.326
Sauces		0.697
Fruit		0.562
Nuts	0.206	0.519
Organ meat		0.513	0.233
Simple sugar	0.289	0.470
Snacks	0.323	0.413
High fat dairy products	0.263	0.345	0.338
Vegetables			0.665
Egg		−0.264	0.568
Red meat	0.220		0.516
Tea and coffee	0.228		0.502
Legumes			0.487
Refined grains			0.301

a

Factor loadings of less than 0.2 have been omitted for simplicity.

Western Dietary Pattern

The Western Pattern included a high intake of chicken, fish, fast foods, low-fat dairy products, olives and olive oil, and hydrogenated oils. It was characterized by urban and modern dietary habits influenced by globalization. The food groups with the highest positive loadings in this pattern were chicken (λ = −0.671), whole grains (λ = 0.647), and flavorings (λ = 0.639).

This pattern accounted for a considerable proportion of variance in dietary intake, suggesting its prevalence among younger, urbanized patients with greater access to processed and convenience foods.

Mixed Dietary Pattern

The Mixed Pattern reflected a diverse combination of both healthy and less healthy food choices. Key contributors included fruits (λ = 0.562), nuts (λ = 0.519), organ meats (λ = 0.513), and simple sugars (λ = 0.470). This pattern highlights the transitional dietary behaviors observed in the study population.

Patients adhering to this pattern often exhibited variability in nutrient intake, with higher consumption of both micronutrient-rich and calorie-dense foods.

Traditional Dietary Pattern

The Traditional Pattern was strongly linked to local cultural practices and traditional cooking methods. It featured high intake of vegetables (λ = 0.665), eggs (λ = 0.568), red meat (λ = 0.516), legumes (λ = 0.487), and tea and coffee (λ = 0.502). This pattern represents reliance on staple foods commonly consumed in the region.

This pattern was more common among older patients and those from lower socio-economic backgrounds, reflecting the influence of cultural and economic factors on dietary habits.

Associations Between Dietary Patterns and Anthropometric Indices

Tables 4 to 6 present detailed associations between dietary patterns and anthropometric indices after adjustment for confounders such as energy intake, age, sex, physical activity, and socio-economic factors.

Table 4.

The Association Between Western Dietary Pattern (Independent Variable) With Anthropometric Indices and Metabolic Parameters (Dependent Variables) in Hemodialysis Patients.

Variables	Model 1 a			Model 2 b			Model 3 c
	B (Unstandardized)	SE	P-value	B (Unstandardized)	SE	P-value	B (Unstandardized)	SE	P-value
Weight (kg)	5.95	0.74	<.001	4.29	0.83	<.001	−0.57	0.56	.30
BMI (kg/m2)	1.79	0.24	<.001	1.41	0.27	<.001	0.17	0.13	.18
WC (cm)	4.75	0.64	<.001	3.67	0.73	<.001	0.28	0.33	.40
HC (cm)	2.99	0.46	<.001	2.47	0.52	<.001	0.39	0.22	.07
WHR	0.02	0.00	<.001	0.01	0.00	.002	0.00	0.00	.74
WHtR	0.02	0.00	<.001	0.02	0.00	<.001	0.00	0.00	.06
Wrist circumference (cm)	1.04	0.18	<.001	0.71	0.21	.001	−0.07	0.08	.40
Neck circumference (cm)	1.47	0.20	<.001	1.12	0.23	<.001	0.09	0.13	.45
Body frame size	−0.41	0.08	<.001	−0.30	0.09	.002	−0.04	0.05	.46
FBS (mg/dL)	2.24	2.24	.31	0.83	2.57	.03	1.94	2.12	.36
TG (mg/dL)	−0.24	2.73	.92	−2.13	0.49	<.001	−3.53	3.80	.35
TC (mg/dL)	19.34	1.65	<.001	20.97	1.90	<.001	11.81	1.81	<.001
LDL-c (mg/dL)	14.70	1.19	<.001	15.98	1.36	<.001	9.54	1.33	<.001
HDL-c (mg/dL)	−1.29	0.66	.053	−1.27	0.76	.09	−1.64	0.92	.07
LDL-c/HDL.c ratio	0.22	0.02	<.001	0.24	0.02	<.001	0.16	0.03	<.001
AIP	0.11	0.008	<.001	0.11	0.01	<.001	0.08	0.01	<.001
TC/HDL ratio	0.31	0.03	<.001	0.34	0.03	<.001	0.22	0.04	<.001
B.SBP (mmHg)	6.92	1.02	<.001	5.33	1.16	<.001	0.34	0.36	.35
A.SBP (mmHg)	6.36	0.92	<.001	5.03	1.05	<.001	0.38	0.44	.38
ΔSBP	0.55	0.25	.03	0.29	0.29	.31	−0.04	0.26	.86
B.DBP (mmHg)	2.97	0.53	<.001	2.13	0.60	<.001	0.11	0.32	.71
A.DBP (mmHg)	2.83	0.45	<.001	2.17	0.52	<.001	0.27	0.22	.22
ΔDBP	0.14	0.19	.45	−0.03	0.22	.86	−0.15	0.19	.41
B.PP (mmHg)	3.94	0.56	<.001	3.20	0.64	<.001	0.22	0.30	.47
A.PP (mmHg)	3.53	0.51	<.001	2.86	0.58	<.001	0.11	0.31	.71
ΔPP	0.40	0.26	.12	0.33	0.30	.27	0.10	0.31	.73
B.MAP (mmHg)	4.29	0.68	<.001	3.20	0.77	<.001	0.19	0.30	.53
A.MAP (mmHg)	4.01	0.60	<.001	3.12	0.68	<.001	0.31	0.27	.26
ΔMAP	0.28	0.17	.11	0.07	0.20	.71	−0.11	0.16	.46
B.BUN (mg/dL)	6.21	0.91	<.001	4.98	1.04	<.001	1.17	0.78	.13
A.BUN (mg/dL)	1.22	0.17	<.001	0.94	0.20	<.001	0.02	0.11	.85
ΔBUN = B.BUN-A.BUN	4.99	0.76	<.001	4.04	0.87	<.001	1.15	0.71	.10
Plasma Cr (mg/dL)	0.55	0.08	<.001	0.40	0.09	<.001	0.004	0.057	.94
Uric acid (mg/dL)	0.58	0.08	<.001	0.43	0.09	<.001	0.02	0.04	.61
eGFR	0.40	0.12	.001	0.25	0.14	.07	−0.07	0.09	.43
Dialysis adequacy	0.29	0.14	.04	0.30	0.17	.07	0.33	0.19	.08
Na (mEq/L)	1.11	0.16	<.001	0.85	0.18	<.001	0.13	0.10	.192
K (mEq/L)	0.22	0.03	<.001	0.17	0.03	<.001	0.004	0.01	.76
Ca (mg/dL)	0.13	0.02	<.001	0.10	0.02	<.001	−0.003	0.007	.70
P (mg/dL)	0.29	0.04	<.001	0.23	0.04	<.001	0.01	0.01	.37
AST (IU/L)	−0.10	0.51	.83	−0.52	0.59	.37	−1.03	0.71	.14
ALT (IU/L)	0.12	0.51	.80	−0.18	0.58	.75	−0.53	0.71	.45
ALT/AST	0.009	0.008	.25	0.01	0.009	.10	0.02	0.01	.01

Abbreviations: AIP, atherogenic index of plasma; BMI, body mass index; FBS, fasting blood sugar; HC, hip circumference; HDL-c, high-density lipoprotein cholesterol; LDL-c, low-density lipoprotein cholesterol; TC, total cholesterol; TG, triglyceride; WC, waist circumference; WHR, waist-to-hip ratio; WHtR, waist-to-height ratio.

P < .05 was considered as significant.

a

Model 1: linear regression analysis without adjustment.

b

Model 2: linear regression analysis with adjustment for energy.

c

Model 3: linear regression analysis with correction for energy, age, sex, physical activity, race, job, marital status, education, BMI, duration of dialysis, chronic diseases, and medications.

Table 5.

The Association Between Mixed Dietary Pattern (Independent Variable) With Anthropometric Indices and Metabolic Parameters (Dependent Variables) in Hemodialysis Patients.

Variables	Model 1 a			Model 2 b			Model 3 c
	B (Unstandardized)	SE	P-value	B (Unstandardized)	SE	P-value	B (Unstandardized)	SE	P-value
Weight (kg)	0.60	0.81	.45	−5.09	0.94	<.001	−1.58	0.61	.01
BMI (kg/m2)	0.31	0.26	.22	−1.11	0.31	.001	−0.21	0.14	.14
WC (cm)	0.69	0.69	.32	−3.30	0.84	<.001	−0.39	0.36	.27
HC (cm)	0.52	0.49	.28	−1.64	0.61	.008	−0.42	0.24	.08
WHR	0.02	0.04	.60	−0.18	0.05	.001	0.00	0.01	.98
WHtR	0.005	0.004	.26	−0.01	0.006	.01	0.001	0.003	.83
Wrist circumference (cm)	0.14	0.19	.47	−0.89	0.24	<.001	−0.03	0.09	.75
Neck circumference (cm)	0.26	0.22	.23	−0.96	0.27	<.001	−0.20	0.14	.16
Body frame size	−0.74	0.08	.39	0.29	0.11	.007	−0.03	0.06	.57
FBS (mg/dL)	−0.82	2.24	.714	−6.36	2.91	.03	−3.94	2.31	.09
TG (mg/dL)	5.72	2.71	.03	6.83	3.57	.05	9.10	4.12	.02
TC (mg/dL)	−5.13	1.97	.01	−16.89	2.38	<.001	−6.13	2.10	.004
LDL-c (mg/dL)	−3.87	1.44	.008	−12.72	1.73	<.001	−5.01	1.56	.001
HDL-c (mg/dL)	−0.33	0.67	.62	0.17	0.88	.84	0.55	1.01	.58
LDL-c/HDL.c ratio	−0.04	0.02	.09	−0.17	0.03	<.001	−0.07	0.03	.05
AIP	−0.008	0.01	.45	−0.07	0.01	<.001	−0.03	0.01	.01
TC/HDL ratio	−0.05	0.03	.10	−0.23	0.04	<.001	−0.09	0.04	.03
B.SBP (mmHg)	1.55	1.09	.15	−3.87	1.36	.005	−0.40	0.40	.31
A.SBP (mmHg)	1.03	0.99	.29	−4.03	1.22	.001	−0.80	0.48	.09
ΔSBP	0.52	0.25	.04	0.16	0.33	.62	0.39	0.29	.17
B.DBP (mmHg)	1.19	0.55	.03	−1.03	0.70	.14	0.31	0.35	.38
A.DBP (mmHg)	0.82	0.48	.09	−1.28	0.60	.03	0.17	0.24	.47
ΔDBP	0.37	0.19	.05	0.25	0.25	.32	0.13	0.20	.52
B.PP (mmHg)	0.36	0.61	.55	−2.83	0.75	<.001	−0.71	0.33	.03
A.PP (mmHg)	0.21	0.54	.70	−2.74	0.67	<.001	−0.98	0.34	.004
ΔPP	0.15	0.26	.56	−0.08	0.34	.80	0.26	0.34	.43
B.MAP (mmHg)	1.31	0.72	.07	−1.98	0.90	.03	0.07	0.33	.83
A.MAP (mmHg)	0.89	0.64	.16	−2.20	0.80	.006	−0.14	0.30	.62
ΔMAP	0.42	0.17	.01	0.22	0.23	.34	0.22	0.17	.21
B.BUN (mg/dL)	1.13	0.98	.25	−3.60	1.22	.003	−0.50	0.85	.55
A.BUN (mg/dL)	0.20	0.19	.29	−0.81	0.23	.001	−0.04	0.12	.69
ΔBUN = B.BUN-A.BUN	0.93	0.81	.256	−2.79	1.02	.007	−0.46	0.78	.55
Plasma Cr (mg/dL)	0.14	0.09	.12	−0.31	0.11	.006	−0.04	0.06	.43
Uric acid (mg/dL)	0.12	0.09	.16	−0.36	0.11	.002	−0.04	0.05	.44
eGFR	0.01	0.12	.88	−0.44	0.16	.007	−0.13	0.10	.19
Dialysis adequacy	0.05	0.14	.71	−0.05	0.19	.77	−0.08	0.21	.71
Na (mEq/L)	0.22	0.17	.19	−0.67	0.21	.002	−0.09	0.11	.40
K (mEq/L)	0.04	0.03	.17	−0.12	0.04	.007	−0.01	0.01	.42
Ca (mg/dL)	0.03	0.02	.13	−0.06	0.03	.03	−0.003	0.008	.73
P (mg/dL)	0.04	0.04	.32	−0.19	0.05	.001	−0.01	0.01	.35
AST (IU/L)	0.55	0.51	.28	0.29	0.67	.66	0.91	0.78	.24
ALT (IU/L)	0.41	0.51	.41	0.12	0.67	.85	0.60	0.78	.44
ALT/AST	−0.008	0.008	.36	−0.008	0.01	.474	−0.01	0.01	.15

 Abbreviations: AIP, atherogenic index of plasma; BMI, body mass index; FBS, fasting blood sugar; HC, hip circumference; HDL-c, high-density lipoprotein cholesterol; LDL-c, low-density lipoprotein cholesterol; TC, total cholesterol; TG, triglyceride; WC, waist circumference; WHR, waist-to-hip ratio; WHtR, waist-to-height ratio.

P < 0.05 was considered as significant.

a

Model 1: linear regression analysis without adjustment.

b

Model 2: linear regression analysis with adjustment for energy.

c

Model 3: linear regression analysis with correction for energy, age, sex, physical activity, race, job, marital status, education, BMI, duration of dialysis, chronic diseases, and medications.

Table 6.

The Association Between Traditional Dietary Pattern (Independent Variable) With Anthropometric Indices and Metabolic Parameters (dependent Variables) in Hemodialysis Patients.

Variables	Model 1 a			Model 2 b			Model 3 c
	B (Unstandardized)	SE	P-value	B (Unstandardized)	SE	P-value	B (Unstandardized)	SE	P-value
Weight (kg)	−1.01	0.81	.21	−3.15	0.78	<.001	0.19	0.47	.67
BMI (kg/m2)	−0.91	0.25	<.001	−1.57	0.24	<.001	−0.27	0.11	.01
WC (cm)	−2.15	0.68	.002	−3.91	0.66	<.001	−0.91	0.27	.001
HC (cm)	−2.03	0.47	<.001	−3.13	0.46	<.001	−0.44	0.19	.02
WHR	−0.003	0.004	.47	−0.01	0.004	.01	−0.005	0.001	<.001
WHtR	−0.01	0.004	<.001	−0.02	0.004	<.001	−0.009	0.002	<.001
Wrist circumference (cm)	−0.08	0.19	.67	−0.47	0.19	.01	−0.02	0.07	.70
Neck circumference (cm)	−0.45	0.22	.04	−0.97	0.21	<.001	0.02	0.11	.79
Body frame size	0.08	0.08	.31	0.24	0.08	.007	0.04	0.04	.33
FBS (mg/dL)	3.26	2.23	.14	2.09	2.36	.39	−0.33	1.80	.83
TG (mg/dL)	−2.60	2.72	.34	−3.95	2.88	.17	−3.89	3.24	.23
TC (mg/dL)	−2.18	1.99	.27	−5.11	2.05	.01	−0.32	1.65	.84
LDL-c (mg/dL)	−2.75	1.45	.059	−5.09	1.49	.001	−1.67	1.23	.17
HDL-c (mg/dL)	1.24	0.66	.06	1.64	0.70	.02	1.73	0.78	.02
LDL-c/HDL.c ratio	−0.07	0.02	.007	−0.11	0.02	<.001	−0.07	0.02	.01
AIP	−0.02	0.01	.02	−0.04	0.01	<.001	−0.02	0.01	.006
TC/HDL ratio	−0.09	0.03	.01	−0.15	0.03	<.001	−0.09	0.03	.01
B.SBP (mmHg)	−4.26	1.07	<.001	−6.97	1.03	<.001	−0.18	0.15	.23
A.SBP (mmHg)	−3.57	0.97	<.001	−5.94	0.94	<.001	0.29	0.19	.12
ΔSBP	−0.69	0.25	.007	−1.02	0.26	<.001	−0.47	0.22	.03
B.DBP (mmHg)	−1.95	0.54	<.001	−3.23	0.53	<.001	0.11	0.23	.61
A.DBP (mmHg)	−1.59	0.47	.001	−2.69	0.46	<.001	0.09	0.14	.50
ΔDBP	−0.36	0.19	.06	−0.53	0.20	.009	0.02	0.16	.89
B.PP (mmHg)	−2.30	0.59	<.001	−3.74	0.58	<.001	−0.30	0.23	.19
A.PP (mmHg)	−1.97	0.53	<.001	−3.25	0.52	<.001	0.19	0.18	.29
ΔPP	−0.33	0.26	.20	−0.49	0.27	.08	−0.500	0.25	.48
B.MAP (mmHg)	−2.72	0.71	<.001	−4.47	0.68	<.001	0.01	0.17	.92
A.MAP (mmHg)	−2.25	0.63	<.001	−3.77	0.61	<.001	0.16	0.13	.23
ΔMAP	−0.47	0.17	.008	−0.70	0.18	<.001	−0.14	0.13	.30
B.BUN (mg/dL)	−3.05	0.96	.002	−5.27	0.95	<.001	−0.40	0.46	.37
A.BUN (mg/dL)	−0.51	0.18	.006	−0.96	0.18	<.001	0.02	0.07	.77
ΔBUN = B.BUN-A.BUN	−2.53	0.80	.002	−4.31	0.79	<.001	−0.43	0.43	.31
Plasma Cr (mg/dL)	−0.18	0.09	.04	−0.39	0.09	<.001	0.08	0.04	.08
Uric acid (mg/dL)	−0.32	0.09	<.001	−0.55	0.08	<.001	−0.01	0.03	.75
eGFR	0.23	0.12	.06	0.10	0.13	.41	−0.01	0.07	.85
Dialysis adequacy	−0.25	0.14	.09	−0.33	0.15	.03	−0.17	0.15	.25
Na (mEq/L)	−0.42	0.17	.015	−0.82	0.17	<.001	−0.08	0.04	.050
K (mEq/L)	−0.13	0.03	<.001	−0.21	0.03	<.001	0.001	0.005	.89
Ca (mg/dL)	−0.10	0.02	<.001	−0.15	0.02	<.001	−0.002	0.005	.68
P (mg/dL)	−0.15	0.04	<.001	−0.26	0.04	<.001	−0.005	0.007	.48
AST (IU/L)	−0.48	0.51	.34	−0.76	0.54	.16	−0.80	0.61	.19
ALT (IU/L)	−0.38	0.51	.45	−0.63	0.54	.24	−0.56	0.61	.36
ALT/AST	0.008	0.008	.30	0.01	0.009	.19	0.01	0.01	.30

Abbreviations: AIP, atherogenic index of plasma; BMI, body mass index; HC, hip circumference; FBS, fasting blood sugar; LDL-c, low-density lipoprotein cholesterol; HDL-c, high-density lipoprotein cholesterol; TC, total cholesterol; TG, triglyceride; WC, waist circumference; WHR, waist-to-hip ratio; WHtR, waist-to-height ratio.

P < .05 was considered as significant.

a

Model 1: linear regression analysis without adjustment.

b

Model 2: linear regression analysis with adjustment for energy.

c

Model 3: linear regression analysis with correction for energy, age, sex, physical activity, race, job, marital status, education, BMI, duration of dialysis, chronic diseases, and medications.

Western Dietary Pattern

The Western Pattern was positively associated with weight (β = 4.29, SE = 0.83, P < .001), BMI (β = 1.41, SE = 0.27, P < .001), and WC (β = 3.67, SE = 0.73, P < .001). These associations persisted in fully adjusted models, suggesting that adherence to this pattern contributes to increased body weight and central obesity risk.

Additional analysis showed a trend toward increased neck and wrist circumference among participants adhering to this pattern (β = 1.12, P < .001; β = .71, P = .001, respectively). These findings highlight the potential role of Western and processed foods in exacerbating obesity markers.

Mixed Dietary Pattern

The Mixed Pattern showed significant negative associations with weight (β = −5.09, SE = 0.94, P < .001) and WC (β = −3.30, SE = 0.84, P < .001) in adjusted models. Furthermore, this pattern was linked to reduced hip circumference (β = −1.64, P = .008) and improvements in WHR and WHtR metrics in partially adjusted models.

1. Stratified analysis indicated that adherence to this pattern was more strongly associated with improved anthropometric profiles among younger patients, possibly due to better metabolic adaptability in this age group.

Traditional Dietary Pattern

The Traditional Pattern was negatively associated with BMI (β = −1.57, SE = 0.24, P < .001), WC (β = −3.91, SE = 0.66, P < .001), and WHR (β = −.005, SE = 0.001, P < .001). Additionally, this pattern was linked to decreased neck circumference (β = −.97, P < .001), highlighting its potential protective effects against central and general obesity.

Subgroup analysis revealed stronger protective associations in female patients and those with longer dialysis durations, underscoring the potential benefits of culturally traditional diets in specific populations.

Associations Between Dietary Patterns and Metabolic Parameters

The relationships between dietary patterns and metabolic parameters, including glycemic and lipid profiles, blood pressure, and renal and hepatic markers, are detailed in Tables 4 to 6.

Western Dietary Pattern

The Western Pattern was significantly associated with adverse metabolic outcomes. It was positively associated with total cholesterol (TC) (β = 11.81, SE = 1.81, P < .001), LDL-C (β = 9.54, SE = 1.33, P < .001), and atherogenic index of plasma (AIP) (β = .08, SE = 0.01, P < .001). Associations with SBP and DBP were significant in unadjusted models but lost significance after full adjustment. Higher sodium and potassium levels were also observed in this group (P < .001).

Mixed Dietary Pattern

The Mixed Pattern showed protective associations with lipid parameters, including TC (β = −6.13, SE = 2.10, P = .004), LDL-C (β = −5.01, SE = 1.56, P = .001), and AIP (β = −.03, SE = 0.01, P = .01). Significant reductions were also observed in SBP (β = −4.03, SE = 1.22, P = .001) and DBP (β = −1.28, SE = 0.60, P = .03).

Improved markers of renal function, including lower uric acid levels and plasma creatinine, were observed among adherents to this pattern, particularly in partially adjusted models.

Traditional Dietary Pattern

The Traditional Pattern was positively associated with HDL-C (β = 1.73, SE = 0.78, P = .02) and inversely associated with TC/HDL ratio (β = −.09, SE = 0.03, P = .01) and AIP (β = −.02, SE = 0.01, P = .006). Additionally, this pattern was associated with lower SBP (β = −3.57, SE = 0.97, P < .001) and improvements in estimated glomerular filtration rate (eGFR).

Patients following this pattern also exhibited reduced markers of hepatic stress, such as lower aspartate transaminase (AST) levels, suggesting potential hepatic benefits.

Risk of General and Central Obesity by Dietary Pattern Scores

The association between dietary pattern scores and obesity risk is summarized in Table 7.

Table 7.

Odds Ratio (95% CI) for Risk of General and Central Obesity (dependent Variables) According to the Quartiles of Dietary Pattern Scores (Independent Variable) in Hemodialysis Patients.

Variable	Or (CI)	B	P-value *
Western dietary pattern
General obesity
Model 1 a	1.71 (1.28-2.28)	0.53	<.001
Model 2 b	1.62 (1.18-2.22)	0.48	.003
Model 3 c	0.33 (0.07-1.42)	−1.09	.13
Central obesity according to WC
Model 1 a	1.70 (1.30-2.24)	0.53	<.001
Model 2 b	1.56 (1.15-2.10)	0.44	.003
Model 3 c	0.67 (0.17-2.65)	−0.39	.57
Central obesity according to WHR
Model 1 a	1.42 (1.13-1.83)	0.36	.003
Model 2 b	1.38 (1.06-1.79)	0.32	.01
Model 3 c	1.17 (0.65-2.09)	0.16	.59
Central obesity according to WHtR
Model 1 a	1.97 (1.52-2.56)	0.68	<.001
Model 2 b	1.83 (1.38-2.43)	0.60	<.001
Model 3 c	0.97 (0.59-1.59)	−0.02	.91
Mixed dietary pattern
General obesity
Model 1 a	0.96 (0.74-1.25)	−0.03	.78
Model 2 b	0.71 (0.51-0.99)	−0.33	.04
Model 3 c	0.06 (0.00-12.61)	−2.79	.30
Central obesity according to WC
Model 1 a	1.02 (0.79-1.31)	0.02	.84
Model 2 b	0.73 (0.53-0.99)	−0.31	.04
Model 3 c	0.56 (0.15-2.14)	−0.56	.40
Central obesity according to WHR
Model 1 a	1.14 (0.90-1.44)	0.13	.26
Model 2 b	0.99 (0.74-1.31)	−0.01	.94
Model 3 c	0.91 (0.46-1.79)	−0.09	.79
Central obesity according to WHtR
Model 1 a	0.95 (0.75-1.19)	−0.04	.68
Model 2 b	0.63 (0.46-0.85)	−0.46	.003
Model 3 c	0.78 (0.43-1.41)	−0.24	.41
Traditional dietary pattern
General obesity
Model 1 a	0.57 (0.42-0.76)	−0.55	<.001
Model 2 b	0.48 (0.35-0.65)	−0.73	<.001
Model 3 c	1.11 (0.13-9.37)	0.10	.92
Central obesity according to WC
Model 1 a	0.60 (0.46-0.79)	−0.49	<.001
Model 2 b	0.49 (0.36-0.66)	−0.70	<.001
Model 3 c	0.01 (0.00-7.09)	−3.95	.01
Central obesity according to WHR
Model 1 a	0.72 (0.57-0.92)	−0.32	.008
Model 2 b	0.64 (0.49-0.83)	−0.44	.001
Model 3 c	1.45 (0.82-2.56)	0.37	.19
Central obesity according to WHtR
Model 1 a	0.66 (0.52-0.84)	−0.40	.001
Model 2 b	0.53 (0.40-0.70)	−0.62	<.001
Model 3 c	0.83 (0.51-1.34)	−0.18	<.001

Abbreviations: BMI, body mass index; WC, waist circumference; WHR, waist-to-hip ratio; WHtR, waist-to-height ratio.

Central obesity: WC⩾88 cm for female and WC⩾102 cm for male, WHtR ⩾ 0.5 for people under 40 years old and WHtR ⩾ 0.6 for people over 40 years old, and WHR ⩾ 0.8 for female and WHR ⩾ 1 for male.

General obesity: normal (BMI<30) or obese (BMI⩾30).

a

Model 1: unadjusted.

b

Model 2: adjusted for energy.

c

Model 3: adjusted for energy, age, sex, physical activity, race, job, marital status, education, duration of dialysis, chronic diseases, and medications.

*

P < .05 statistically significant by Multivariable Logistic Regression.

Western Pattern

This pattern was associated with higher odds of central obesity based on WC in adjusted models (OR = 1.56, 95% CI: 1.15-2.10, P = .003). However, no significant associations were observed in fully adjusted models (Model 3).

Mixed Pattern

Adherence to the Mixed Pattern was associated with reduced odds of central obesity based on WHtR (OR = 0.63, 95% CI: 0.46-0.85, P = .003) in adjusted models. Participants in the highest quartile of adherence exhibited significant protective effects against both general and central obesity.

Traditional Pattern

The Traditional Pattern was associated with significantly lower odds of general obesity (OR = 0.48, 95% CI: 0.35-0.65, P < .001) and central obesity based on WC (OR = 0.49, 95% CI: 0.36-0.66, P < .001). These associations persisted in fully adjusted models, emphasizing the protective role of this pattern.

Subgroup analysis indicated that the Traditional Pattern offered greater protective benefits for older patients and those with higher dialysis vintage.

Discussion

The central premise of dietary pattern research is that foods are not consumed in isolation; their combined, synergistic effects often exceed the sum of individual nutrient actions.^7,9 In the metabolically fragile hemodialysis population, this holistic approach is particularly relevant because multiple pathways inflammation, oxidative stress, acid–base balance, and hemodynamic load interact simultaneously. Our a posteriori analysis revealed 3 culturally grounded patterns whose associations with anthropometric and metabolic outcomes provide compelling evidence for this theoretical framework.

• Argument 1: The Western pattern as a driver of cardiometabolic deterioration Greater adherence to the Western pattern independently predicted higher BMI (+1.41 kg/m²), waist circumference (+3.67 cm), total cholesterol (+11.81 mg/dL), LDL-C (+9.54 mg/dl), and atherogenic index a cluster that mirrors the metabolic syndrome phenotype repeatedly described in both general and CKD cohorts exposed to processed-food-dominant diets.^29,30,31 This is not merely an observational correlation. High sodium and refined carbohydrate content activates the renin–angiotensin–aldosterone system and sympathetic nervous system, promoting salt-sensitive hypertension and fluid retention pathways already upregulated in dialysis patients.^32,33,34,35 Concurrently, trans fats and AGEs abundant in processed meats and fast foods impair insulin receptor signaling and increase hepatic VLDL secretion via SREBP-1c and ChREBP activation.^36,37,38 The resulting dyslipidemia and central adiposity amplify cardiovascular risk through established mechanisms: oxidized LDL-induced endothelial dysfunction, macrophage foam-cell formation, and plaque instability.^30,39,40,41 The magnitude of LDL-C elevation observed here is clinically meaningful; in dialysis trials, every 10 mg/dL increase in LDL-C is associated with a 12% to 15% higher risk of cardiovascular events (Figure 2). ⁴² Thus, the Western pattern appears to act as an accelerant of residual cardiovascular risk in an already high-risk population.

• Argument 2: The Mixed pattern reflects nutritional transition and competing biological forces The Mixed pattern embodies the ongoing nutrition transition in Iran: coexistence of traditional plant-based foods with modern energy-dense, nutrient-poor items. ⁴³ Its divergent associations protective against central obesity, LDL-C, and uric acid, yet harmful for triglycerides are biologically plausible and illustrate competing pathways. Antioxidant polyphenols from fruits and nuts scavenge reactive oxygen species and upregulate Nrf2-mediated defenses, while soluble fiber slows gastric emptying and reduces postprandial lipemia.^44,45 In contrast, simple sugars trigger rapid insulin spikes, activating hepatic de novo lipogenesis and ApoC-III expression, thereby inhibiting lipoprotein lipase and raising triglycerides.^38,46 The observed reduction in serum uric acid (−0.75 mg/dL per SD) likely reflects lower fructose-driven purine metabolism and reduced net endogenous acid production an effect with direct renoprotective implications, as hyperuricemia is an independent predictor of dialysis graft loss and mortality (Figure 3).^47,48 This pattern therefore represents a transitional state where protective traditional elements partially buffer, but do not fully neutralize, emerging metabolic threats.

• Argument 3: The Traditional pattern approximates an evidence-based “renal-protective” diet within cultural boundaries The strongest and most consistent beneficial associations were observed with the Traditional pattern, characterized by vegetables, legumes, eggs, moderate red meat, and tea/coffee. This empirically derived pattern closely resembles the plant-based, low-acid-load diets shown in randomized trials to slow eGFR decline, reduce inflammation, and improve survival in CKD.^11,49,50 Several converging mechanisms explain its superiority:

• ● High dietary fiber and polyphenols enhance satiety via GLP-1 secretion, reduce energy harvest, and increase fecal bile acid excretion, lowering LDL-C and body weight^51,52;

• ● Alkali-generating foods (vegetables, legumes) decrease net endogenous acid production, mitigating metabolic acidosis-induced muscle wasting and bone disease^53,54;

• ● Potassium- and magnesium-rich components counteract sodium-mediated vasoconstriction and lower blood pressure through natriuresis and direct vasodilatory effects^55,56,57;

• ● Polyphenols and nitrate from vegetables upregulate endothelial nitric oxide synthase, improving arterial compliance. ⁵⁸ The observed 3.57 mmHg reduction in systolic blood pressure is clinically significant; in hemodialysis cohorts, each 10 mmHg lower SBP is associated with a 15% to 20% reduction in cardiovascular mortality (Figure 4). ⁵⁹ Importantly, because this pattern emerged naturally from local eating habits, it offers a culturally acceptable template for dietary counseling a critical advantage over imported Mediterranean or DASH prescriptions that often face adherence barriers in Middle Eastern settings.

Figure 2.

Mechanism of Western diet on metabolic indices. Mechanistic pathway linking Western dietary pattern to cardiovascular risk in hemodialysis patients (Model 3, fully adjusted, n = 300). The Western pattern, characterized by high intake of processed meats, fast food, hydrogenated oils, and refined grains, demonstrates significant associations with adverse lipid profiles after full adjustment for energy intake, age, sex, physical activity, race, occupation, marital status, education, BMI, dialysis duration, comorbidities, and medications. Six significant harmful effects were observed: elevated total cholesterol (+11.81 mg/dL, P < .001), LDL cholesterol (+9.54 mg/dL, P < .001), LDL/HDL ratio (+0.16, P < .001), atherogenic index of plasma (+0.08, P < .001), TC/ HDL ratio (+0.22, P < .001), and ALT/AST ratio (+0.02, P = .01). These effects are mediated through upregulation of HMGCoA reductase, downregulation of LDL receptors, trans fat-induced insulin resistance, advanced glycation end-products (AGEs),increased hepatic VLDL secretion, chronic inflammation, and endothelial dysfunction. Dyslipidemia emerged as the primary independent mechanism for increased cardiovascular risk, while anthropometric and blood pressure effects lost significance after full adjustment, suggesting mediation through energy intake and BMI confounders. This pattern should be avoided in hemodialysis patients due to its strong atherogenic potential.

Figure 3.

Mechanism of mixed diet on metabolic indices. Mechanistic pathway illustrating the dual metabolic effects of the Mixed dietary pattern in hemodialysis patients (Model 3, fully adjusted, n = 300). The Mixed pattern, characterized by coexistence of protective foods (fruits, nuts, organ meats) and harmful components (simple sugars, sweets, high-fat dairy, salty snacks), reflects the ongoing nutrition transition in Middle Eastern populations. After full adjustment for confounders, this pattern demonstrated 6 significant effects: 5 beneficial outcomes including reduced weight (−−1.58 kg, P = .01), total cholesterol (−6.13 mg/dL, P = .004), LDL cholesterol (−5.01 mg/dL, P = .001), atherogenic index (−0.03, P = .01), and TC/HDL ratio (−0.09, P = .03); alongside 1 harmful effect of elevated triglycerides (+9.10 mg/dL, P = .02). These competing effects are mediated through protective mechanisms (antioxidants reducing oxidative stress, fiber decreasing LDL absorption, polyphenols providing anti-inflammatory effects) counterbalanced by harmful pathways (simple sugars triggering hepatic lipogenesis and increased triglyceride synthesis via ApoC-III expression). The net result represents partial cardioprotection with residual hypertriglyceridemia. Clinical modification is recommended: maintaining consumption of protective foods (fruits, nuts) while limiting simple sugar intake to optimize cardiovascular benefits in this high-risk population.

Figure 4.

Mechanism of traditional diet on metabolic indices. Mechanistic pathway demonstrating comprehensive cardioprotective effects of the Traditional dietary pattern in hemodialysis patients (Model 3, fully adjusted, n = 300). The Traditional pattern, characterized by high intake of vegetables, legumes, eggs, red meat, tea/coffee, high-fat dairy, and refined grains with minimal processed foods, showed the strongest and most comprehensive beneficial associations among all 3 dietary patterns. After full adjustment for confounders, 10 significant protective effects were observed across multiple domains: (1) Anthropometric improvements (n = 5): reduced BMI (−0.27 kg/m², P = .01), waist circumference (−0.91 cm, P = .001), hip circumference (−0.44 cm, P = .02), waist-to-hip ratio (−0.005, P < .001), and waist-to-height ratio (−0.009, P < .001); (2) Favorable lipid profile changes (n = 4): increased HDL cholesterol (+1.73 mg/dL, P = .02)—the only pattern showing HDL elevation—along with reduced LDL/HDL ratio (−0.07, P = .01), atherogenic index (−0.02, P = .006), and TC/HDL ratio (−0.09, P = .01); (3) Blood pressure improvement (n = 1): reduced systolic blood pressure change (ΔSBP: −0.47 mmHg, P = .03). These effects are mediated through multiple protective biological mechanisms including high dietary fiber increasing satiety and reducing LDL via bile acid excretion, polyphenols from vegetables and tea decreasing oxidative stress and improving arterial compliance via eNOS upregulation, potassium and magnesium promoting natriuresis and vasodilation, alkali-generating foods reducing metabolic acidosis, and plant proteins providing anti-inflammatory effects. This culturally-rooted, plant-predominant pattern offers the most comprehensive cardiovascular protection and should be actively promoted as a first-line dietary intervention for hemodialysis patients in Middle Eastern settings due to its cultural acceptability, sustainability, and evidence-based multi-domain benefits.

The interconnected mechanisms by which these 3 dietary patterns influence metabolic, anthropometric, and clinical outcomes are summarized in Figure 5. As illustrated, the Western pattern drives a cascade of pro-inflammatory, pro-oxidant, and acidotic stimuli that converge on increased cardiovascular and renal stress. In contrast, the Traditional pattern activates opposing anti-inflammatory, antioxidant, and alkali-generating pathways, while the Mixed pattern occupies an intermediate position with partial neutralization of harmful effects. This schematic underscores the complexity and clinical relevance of food-synergy-based dietary impacts in hemodialysis patients.

• Clinical implications: The magnitude of the observed associations has direct clinical relevance for hemodialysis patients, a population with >50% 5-year cardiovascular mortality. Adherence to the Traditional pattern was associated with a 1.57 kg/m² lower BMI, 3.91 cm smaller waist circumference, 3.57 mmHg lower systolic blood pressure, and a 1.73 mg/dL higher HDL-C changes that, if achieved population-wide, could translate into substantial risk reduction. For example, a 3 to 4 mmHg reduction in SBP is estimated to reduce cardiovascular events by 10% to 15% in dialysis cohorts, ⁵⁹ while each 1 kg/m² lower BMI is linked to 8% to 12% lower all-cause mortality. ⁶⁰ Conversely, each standard deviation increase in Western pattern score raised LDL-C by 9.54 mg/dL an increment comparable to the effect of missing statin therapy in this high-risk group. ⁴² These effect sizes suggest that shifting just 20% to 30% of patients from Western/Mixed toward Traditional eating habits through culturally tailored counseling could yield clinically meaningful improvements in cardiometabolic profile, potentially reducing cardiovascular events and hospitalizations without additional pharmacological burden. This is particularly feasible in Middle Eastern settings where many components of the Traditional pattern (legumes, vegetables, tea) are already affordable and culturally acceptable.

Figure 5.

Mechanisms of dietary patterns. Comparative mechanistic pathways of 3 dietary patterns identified through principal component analysis in hemodialysis patients (Model 3, fully adjusted, n = 300). This figure synthesizes the distinct metabolic effects, underlying mechanisms, and cardiovascular outcomes of Western, Mixed, and Traditional dietary patterns. The Western pattern (left panel, red) demonstrated 6 significant harmful effects, exclusively affecting lipid metabolism: elevated total cholesterol (+11.81 mg/dL), LDL cholesterol (+9.54 mg/dL), atherogenic indices, and hepatic stress markers (all P ⩽ .01), mediated through pro-atherogenic mechanisms including upregulated cholesterol synthesis, downregulated LDL receptors, and chronic inflammation, resulting in increased cardiovascular risk with dyslipidemia as the primary pathway. The Mixed pattern (middle panel, purple) exhibited competing biological forces with 5 beneficial effects (reduced weight, total cholesterol, LDL cholesterol, and atherogenic indices) counterbalanced by 1 harmful effect (elevated triglycerides + 9.10 mg/dL, P = .02), reflecting the nutrition transition where protective antioxidants and fiber partially neutralize the lipogenic effects of simple sugars, yielding partial cardioprotection requiring dietary modification. The Traditional pattern (right panel, green) demonstrated superior comprehensive protection with 10 significant beneficial effects across anthropometric (n = 5), lipid (n = 4), and blood pressure (n = 1) domains, mediated through multiple protective pathways including fiber, polyphenols, alkalizing foods, and minerals, resulting in the strongest reduction in cardiovascular risk. Clinical hierarchy: Traditional (10 benefits) > Mixed (net 4 benefits) > Western (6 harms). These findings support a priority intervention strategy of shifting hemodialysis patients from Western toward Traditional dietary patterns through culturally-adapted nutrition counseling, leveraging the Traditional pattern’s alignment with Middle Eastern culinary heritage to achieve sustainable adherence and optimal cardiovascular outcomes in this high-risk population with >50 % five-year mortality.

In summary, our findings provide robust, mechanism-based evidence that dietary patterns are powerful modulators of cardiometabolic health in hemodialysis patients. The Traditional Iranian pattern aligns closely with theoretical constructs of anti-inflammatory, alkali-generating, plant-predominant diets and confers measurable clinical benefits. Shifting patients away from Western and Mixed patterns toward greater adherence to culturally rooted Traditional eating may represent one of the most feasible and effective non-pharmacological interventions in this high-risk population.

Conclusion

The findings of this study emphasize the critical role of dietary patterns in influencing the metabolic and anthropometric health of hemodialysis patients. Among the dietary patterns identified, the Traditional pattern emerged as a particularly beneficial approach, associated with improved lipid profiles, lower blood pressure, and reduced central obesity. Conversely, the Western pattern was linked to adverse outcomes, such as increased body mass index and elevated cholesterol levels, highlighting the risks associated with modern dietary habits rich in processed foods. These results underscore the necessity of implementing dietary interventions that are both nutritionally balanced and culturally appropriate. Educating patients about the benefits of adhering to traditional diets, while limiting the intake of Western and processed foods, could significantly enhance their overall health and quality of life. Additionally, healthcare providers should consider integrating dietary counseling into routine care for hemodialysis patients, focusing on personalized approaches that account for individual preferences and socio-cultural factors. Future research should explore the long-term effects of these dietary patterns through longitudinal studies, as well as investigate the potential mechanisms underlying the observed associations. Such efforts will provide deeper insights into optimizing dietary strategies for this vulnerable population and contribute to the broader understanding of nutrition in chronic disease management. In clinical practice, these findings support screening hemodialysis patients for dominant dietary patterns using brief PCA-derived tools and prioritizing nutrition counseling that reinforces Traditional Iranian eating (vegetable- and legume-centered meals, minimal processed foods) while explicitly discouraging Western-type items. Such a culturally congruent, pattern-focused approach is likely to achieve higher long-term adherence and better cardiometabolic outcomes than conventional rigid renal diets, offering a practical, low-cost intervention to improve survival and quality of life in this vulnerable population.

Clinically, these findings advocate for:

• Replacing Western-processed foods with Traditional staples (vegetables, legumes, whole grains) in renal dietary guidelines.

• Screening hemodialysis patients for dietary patterns using short PCA-derived tools to identify high-risk profiles.

• Integrating cultural competence into nutritional counseling—for example, promoting locally relevant foods like lentil stew (Iran) over generic “low-sodium” advice. This shift from nutrient-centric to pattern-focused care could improve adherence and clinical outcomes in resource-limited settings.

Strengths and Limitations

This study has several notable strengths. First, it employed a robust analytical approach using factor analysis to identify dietary patterns, which provides a comprehensive understanding of overall diet quality rather than focusing on individual nutrients or food items. Second, the study included a relatively large sample of hemodialysis patients, enhancing the generalizability of the findings to similar populations. Third, the use of validated tools, such as the 168-item Food Frequency Questionnaire (FFQ) and standardized anthropometric and metabolic measurements, ensured the reliability and accuracy of the collected data. These strengths collectively contribute to the study’s contribution to the growing body of literature on nutrition and metabolic health in chronic kidney disease (CKD). However, the study also has limitations. Its cross-sectional design precludes causal inferences, as it captures associations at a single point in time and cannot distinguish whether dietary patterns precede or result from metabolic changes. This is particularly relevant in hemodialysis patients, where poor health outcomes (eg, obesity or dyslipidemia) might influence dietary choices, leading to potential reverse causation bias and limiting our ability to recommend interventions based solely on these findings. To mitigate this, we grounded our hypotheses in evidence from prospective and longitudinal studies^11,29 that support similar directionality, and we explicitly excluded patients with recent dietary changes or participation in weight management programs to reduce the likelihood of reverse causation. Self-reported dietary data, while necessary for large-scale dietary assessments, may introduce recall or reporting bias, potentially affecting the accuracy of dietary intake estimation. For instance, participants might underreport unhealthy foods due to social desirability or overestimate portion sizes for culturally favored items, leading to misclassification of pattern adherence and attenuated associations. To address this, we used a validated 168-item FFQ specifically adapted for Iranian populations, ¹⁶ trained interviewers to probe responses neutrally, excluded cases with implausible energy intakes (<800 or >4200 kcal/day), and adjusted for total energy intake in regression models to account for systematic reporting errors. Additionally, factors such as food preparation methods, cultural nuances, and unmeasured confounders like physical activity levels and socioeconomic status may have influenced the outcomes. Food preparation (eg, frying vs steaming) can alter nutrient bioavailability and glycemic load, while socioeconomic factors might correlate with both pattern adherence and access to healthcare, introducing residual confounding that could bias estimates toward or away from the null. We mitigated this by incorporating proxies for socioeconomic status (eg, education and occupation) in fully adjusted models, including detailed questions on cooking methods within the FFQ, and controlling for physical activity via the validated IPAQ questionnaire. ²² The lack of molecular data also restricts insights into the specific biological mechanisms linking dietary patterns to metabolic and renal health. Without biomarkers like C-reactive protein, interleukin-6, or oxidative stress markers, we cannot directly confirm inflammatory or oxidative pathways, potentially limiting the depth of mechanistic interpretations and the ability to personalize interventions. To overcome this, we interpreted associations using well-established mechanistic evidence from cited experimental and cohort studies (eg, Refs.^36,40,58), and we recommend future studies incorporate such biomarkers for validation. Future research should aim to address these limitations by conducting longitudinal and interventional studies to validate these findings and establish causal relationships. Investigating the role of molecular pathways through which dietary components affect health outcomes will also provide valuable insights. Moreover, tailored dietary recommendations that consider individual and cultural differences in food preferences and preparation methods are essential to optimize the nutritional care of hemodialysis patients.

Key limitations include:

1. Cross-sectional design precludes causal inference between diet and outcomes. Mitigation: Hypotheses were grounded in prospective literature,^11,29 and reverse causation was addressed by excluding recent dietary changers.

2. Self-reported FFQ data risks recall/measurement bias. Mitigation: We used a validated 168-item FFQ, ¹⁶ excluded implausible energy intakes (<800/ > 4200 kcal), and controlled for energy in models.

3. Unmeasured confounders (eg, socioeconomic status, food preparation methods). Mitigation: We adjusted for education/occupation as proxies and included detailed cooking methods in the FFQ.

4. Lack of molecular data (eg, inflammatory cytokines). Mitigation: Associations were interpreted using mechanistic evidence from cited studies (eg, Refs.^36,40,59).