Association of high-fat diet with bone mineral density in older U.S. adults: a cross-sectional study of NHANES 2017–2018 Authors: Na Li, Qing He, Yuan Cheng, Tao Jin Published: 2025 ABSTRACT BackgroundEvidence on the relationship between dietary fat intake and bone mineral density (BMD) remains inconsistent, particularly regarding sex- and site-specific differences in femoral bone health among older adults.MethodsThis cross-sectional study analyzed data from the 2017–2018 National Health and Nutrition Examination Survey (NHANES) to explore associations between dietary fat intake and femoral BMD in U.S. adults aged ≥ 50 years. Femoral BMD at four anatomical sites (total femur, neck, trochanter, intertrochanteric region) was measured using dual-energy X-ray absorptiometry (DXA). A high-fat diet was defined as > 35% of total energy intake from fat (%E). Sex-stratified, survey-weighted multivariable logistic regression models were applied to examine associations with osteopenia/osteoporosis after adjustment for relevant covariates.ResultsAmong 1,127 adults aged ≥ 50 years (612 men; 515 women), high-fat diet consumption was more prevalent in men (59.5%) than in women (54.0%). Women showed a higher baseline prevalence of osteopenia/osteoporosis, approximately 1.7–2.0 times that of men across femoral sites. After multivariable adjustment, a high-fat diet was associated with lower odds of femoral neck osteopenia/osteoporosis in men (OR = 0.45, p = 0.03) and women (OR = 0.43, p = 0.04), with no significant associations observed at other sites. Sensitivity and substitution analyses yielded directionally similar results, suggesting a modest, potentially protective association when carbohydrates were replaced with dietary fat. Restricted cubic spline and exploratory dose–response analyses demonstrated a protective effect of moderate fat intake in both sexes, with distinct sex-specific patterns: a nonlinear trend in women—with maximal protection observed at 40–45%E (OR = 0.22, p = 0.02)—and a linear inverse association in men, with the lowest risk at 35–40%E (OR = 0.35, p = 0.01).ConclusionsThese findings suggest a modest, sex- and site-specific association between dietary fat intake and femoral bone health in older adults. The observed relationships should be interpreted as exploratory and hypothesis-generating, emphasizing the need for longitudinal studies to confirm causality. Maintaining balanced macronutrient intake, rather than strict fat restriction, may be a reasonable dietary strategy for preserving bone health in aging populations. Introduction Osteoporosis, a systemic skeletal disorder characterized by reduced bone mass, microarchitectural deterioration, and increased fracture susceptibility, disproportionately impacts postmenopausal women and men aged ≥ 50 years [1, 2]. Clinical diagnosis of osteoporosis and its precursor osteopenia primarily relies on bone mineral density (BMD) T-scores, with hip fractures—a devastating complication—showing direct correlation with diminished hip BMD [1, 3]. Globally, osteoporosis-related fractures generate healthcare costs exceeding many chronic diseases, and the burden is expected to rise with population aging, posing serious challenges to healthcare systems and societal well-being [4–6]. Conventional prevention strategies have emphasized calcium and vitamin D intake, along with lifestyle modification (e.g., smoking cessation, reduced alcohol consumption, and physical activity) [7]. However, emerging evidence suggests that dietary fat intake may also play a critical role in skeletal health, though its impact remains insufficiently studied and controversial [8–11]. The biological effects of dietary fat on bone homeostasis appear bidirectional. High-fat diets may promote bone loss through oxidative stress, chronic inflammation, and gut microbiota dysbiosis, disrupting osteoblast–osteoclast balance [12, 13]. Conversely, dietary lipids may provide benefits by enhancing weight-bearing mechanical loading and improving absorption of fat-soluble vitamins, particularly D and K, which regulate calcium metabolism and bone mineralization [13–16]. These paradoxical findings highlight the need for more nuanced investigation. Human studies have yielded inconsistent results. For example, Kwon et al. found both low- and high-fat diets were linked to lower BMD in women, without detailed site-specific bone assessments [17]; Corwin et al. reported an inverse association between saturated fat and femoral neck BMD in younger men, but did not examine other hip subregions [18]; Kim et al. showed sex differences, with high fat intake reducing bone strength in men, yet lacked regional bone assessments [19]. These discrepancies likely reflect methodological limitations, particularly the absence of sex-stratified analyses and site-specific evaluations, despite well-known sexual dimorphism in bone metabolism and regional variations in fracture susceptibility [17–20]. To address these uncertainties, we conducted a comprehensive sex-stratified analysis using data from the 2017–2018 NHANES, focusing on dietary fat intake and BMD across multiple femoral sites in U.S. adults aged 50 years and older. This study extends previous research by systematically examining site-specific skeletal responses to fat intake and elucidating potential sex-dependent differences. By targeting a population at high risk for osteoporosis and adjusting for a wide range of demographic, lifestyle, metabolic, and hormonal factors, our analysis aims to provide robust, population-based evidence on the contextual relationship between dietary fat and bone health, thereby informing more tailored nutritional strategies for bone preservation in aging adults. Methods Study population This cross-sectional analysis utilized data from the 2017–2018 National Health and Nutrition Examination Survey (NHANES), a nationally representative survey conducted by the Centers for Disease Control and Prevention (CDC) employing complex multistage probability sampling. A total of 3,069 adults aged 50 years or older were initially included. Then, we excluded participants missing mobile examination center (MEC) data (n = 171) and those without DXA measurements (n = 612). Among those who completed two 24-hour dietary recall interviews, individuals with unreliable dietary recall status or missing covariate data were further excluded. To minimize bias from extreme dietary misreporting, participants with implausible total energy intakes (< 500 kcal/day for all participants, > 5,000 kcal/day for women, or > 8,000 kcal/day for men) were also removed from the analytic sample. To enhance internal validity, only participants responding “usual” to the dietary recall question (“Was the amount of food consumed yesterday much more than usual, usual, or much less than usual?”) were included for macronutrient distribution and energy intake calculation. The final analytical sample included 1,127 participants who reported “usual” dietary intake (Fig. 1). To assess the robustness of the findings, sensitivity analyses were additionally conducted including all eligible participants (n = 1,865), regardless of their response to the “usual” intake question. All NHANES datasets used were fully de-identified, and the protocol was approved by the Institutional Review Board. Fig. 1Flow diagram of the study participants Flow diagram of the study participants Clinical and laboratory data collection Demographic variables included gender (Male, Female), age, race/Hispanic origin (Mexican American, Other Hispanic, Non-Hispanic White, Non-Hispanic Black, and Other Race), and marital status categorized as Married/Living with partner, Widowed/Divorced/Separated, or Never Married. Smoking status was determined through the question “Smoked at least 100 cigarettes in entire life?” with affirmative responses classified as smokers. Alcohol use was defined by positive responses to “Ever had a drink of any kind of alcohol?“. Physical activity assessment included the specific question: “Does your work involve vigorous-intensity activity that causes large increases in breathing or heart rate like carrying or lifting heavy loads, digging or construction work for at least 10 minutes continuously in a typical week?” with “Yes” responses indicating vigorous work activity. Family history of osteoporosis was determined based on participants’ affirmative responses to the NHANES questionnaire item: “Has a doctor ever told either of your parents that they had osteoporosis?“. Female participants who affirmed “Yes” to the question: “Have you ever used female hormones such as estrogen and progesterone?” were classified as having prior hormone exposure. This classification encompassed all administration forms (oral tablets, transdermal creams/patches, and injectable preparations), while explicitly excluding contraceptive methods and infertility treatments. Anthropometric measurements followed NHANES protocols: waist and hip circumferences were measured using steel tape, and body mass index (BMI) was calculated as weight(kg)/height(m)². Abdominal obesity was defined by gender-specific waist circumference thresholds (> 102 cm for men, > 88 cm for women) [21]. Blood pressure (BP) measurements were obtained three times consecutively and averaged. Hypertension diagnosis required systolic blood pressure ≥ 140 mmHg and/or diastolic blood pressure ≥ 90 mmHg or current antihypertensive medication use [22]. Diabetes mellitus was confirmed through any of: (1) self-reported diagnosis; (2) antidiabetic medication use; (3) fasting plasma glucose ≥ 126 mg/dL; (4) glycated hemoglobin (HbA1c) level ≥ 6.5%; (5) random plasma glucose concentration ≥ 200 mg/dL [23]. Laboratory analyses included fasting measurements of high-density lipoprotein-cholesterol (HDL-C), low density lipoprotein-cholesterol (LDL-C), total cholesterol, 25-hydroxyvitamin D3 (25-OHD3), total calcium, blood cadmium and high sensitivity c-reactive protein (HS-CRP). Nutritional biomarkers (total saturated/monounsaturated/polyunsaturated fatty acids) were extracted from the total nutrient intakes file. Bone mineral density measurement BMD (g/cm²) was measured by dual-energy X-ray absorptiometry (DXA) using a Hologic Discovery model (Hologic Inc., Bedford, MA, USA) at four femoral sites: total femur, neck, trochanter, and intertrochanter. NHANES performs daily calibration and cross-site quality control procedures to ensure measurement consistency. Standard scanning protocol prioritized the left hip, except when participants reported left-hip fracture, hip replacement, or internal fixation; in such cases, the right hip was scanned. Exclusion criteria for DXA included pregnancy, contrast agent exposure within 7 days, weight > 450 lb (204.1 kg), or bilateral hip conditions. BMD classification followed the diagnostic criteria established by the World Health Organization (WHO) [24], utilizing T-scores calculated as: (participant’s BMD measurement - mean BMD of reference population)/standard deviation (SD) of reference population. The diagnostic thresholds were defined as: osteopenia for T-scores between − 2.5 and − 1.0, and osteoporosis for T-scores ≤-2.5. The reference population for hip T-score calculations consisted of non-Hispanic white women aged 20–29 years from the NHANES III database [25]. Given that WHO diagnostic thresholds are primarily validated for femoral neck and total hip, the femoral neck BMD served as the primary endpoint, while total, trochanteric, and intertrochanteric BMD were analyzed as secondary exploratory outcomes. Continuous BMD values were additionally analyzed to complement categorical classifications. Dietary intake assessment All participants completed two 24-hour dietary recall interviews: the first conducted in-person at the Mobile Examination Center (MEC) and the second by telephone 3–10 days later. Daily nutrient intakes were averaged across the two recalls to estimate habitual dietary intake, consistent with standard NHANES analytic practice. Total energy intake was calculated as the sum of energy from protein, carbohydrates, total fat, and alcohol using standard conversion factors: 4 kcal/g for protein and carbohydrates, 9 kcal/g for total fat, and 7 kcal/g for alcohol [26, 27]. Macronutrient intake was expressed as a percentage of total energy intake (%E) through energy-adjusted conversion of gram quantities. A high-fat diet was defined as > 35%E from total fat according to the 2015–2020 Dietary Guidelines for Americans, which recommend ≤ 35%E from fat for adults aged ≥ 50 years [28]. This threshold corresponds to the upper limit of the Acceptable Macronutrient Distribution Range (AMDR) proposed by the Institute of Medicine and thus provides a physiologically and policy-relevant cutoff for defining high-fat intake in older adults [29]. To minimize information loss from dichotomization, total dietary fat intake (%E) was also analyzed as a continuous exposure using restricted cubic spline (RCS) models with four knots at the 5th, 35th, 65th, and 95th percentiles. A non-significant global F-test (p > 0.05) indicated an approximately linear association. As an exploratory dose-stratified sensitivity analysis, participants were further categorized by total fat intake thresholds—Group I: 35% < total fat ≤ 40%E; Group II: 40% < total fat ≤ 45%E; and Group III: >45%E—based on the dietary guidelines of the National Academy of Medicine. This stratification allowed examination of risk patterns across different fat intake levels, evaluation of linear trends, assessment of the stability of high-fat associations, and testing of model robustness. In addition, major fat subtypes—saturated fatty acids (SFA), monounsaturated fatty acids (MUFA), and polyunsaturated fatty acids (PUFA)—were examined separately to evaluate differential associations. To evaluate nutrient substitution effects, isocaloric substitution models were implemented to estimate the theoretical effect of replacing 5%E from carbohydrates or protein with an equivalent proportion of total fat, while holding total energy intake constant. This approach follows the energy-partitioning framework and allows assessment of macronutrient substitution under energy-balanced conditions. Statistical analysis All analyses were conducted using Stata 17.0. NHANES survey design variables (two-day dietary recall weights [WTDR2D], strata [SDMVSTRA], and PSUs [SDMVPSU]) were incorporated using the svyset command following NCHS guidance. Categorical variables were summarized as weighted proportions ± SE, and continuous variables as weighted means ± SE or medians (IQR). Between-group differences (high-fat vs. normal-fat diet) were evaluated using survey-weighted t-tests for continuous variables and survey-weighted logistic regression models for categorical variables. Associations between dietary fat and bone outcomes were assessed using survey-weighted regression models, stratified by sex. (1) Primary analysis: Multivariable logistic regression models (svy: logit Disease_nk i.highfat) estimated odds ratios (ORs, 95% CIs) for femoral neck osteopenia/osteoporosis comparing high-fat to normal-fat diets. Three models were sequentially adjusted: Model I: unadjusted; Model II: age, race, total energy intake, smoking, alcohol, vigorous activity, and estrogen replacement therapy; Model III (fully adjusted): Model II covariates plus marital status, HDL-C, LDL-C, total cholesterol, triglycerides, 25-OHD₃, total calcium, HS-CRP, blood cadmium, hypertension, diabetes, and family history of osteoporosis. Body composition variables (BMI, waist/hip circumference, abdominal obesity) were viewed as potential mediators in the adiposity–metabolic–bone pathway, and thus excluded from main models but included in sensitivity analyses [9]. (2) Continuous dose–response: Total fat intake (%E) was analyzed as a continuous exposure (svy: logit Disease_nk c.fat_pct) to obtain ORs per 1%E increment. Nonlinearity was examined using RCS with four knots at the 5th, 35th, 65th, and 95th percentiles; a non-significant global F-test indicated an approximately linear association. (3) Linear validation: To validate the direction of association, survey-weighted linear regressions (svy: regress BMD_nk i.highfat and svy: regress BMD_nk c.fat_pct) estimated the mean difference or β coefficient for BMD per 1%E increase in total fat. (4) Sensitivity and exploratory analyses: Other skeletal sites (total femur, trochanter, intertrochanteric region); Fat subtypes (SFA, MUFA, PUFA) modeled as %E; Isocaloric substitution of 5%E carbohydrate or protein with total fat; High-fat intake subgroups (≤ 35%E as reference); Inclusion of “unusual intake days” to test model robustness. Sex-specific and fat-intake–stratified analyses were considered exploratory and hypothesis-generating, acknowledging limited subgroup sample sizes and statistical power. Missing data (> 10%) were primarily observed for triglycerides and LDL-C, both showing monotone patterns. Multiple imputation was performed using predictive mean matching (five nearest neighbors, m = 20 imputations) under the chained-equation framework adapted for monotone data structures. The imputation model included all exposure, outcome, and covariate variables, along with NHANES survey design elements (weights, strata, and PSU identifiers) to preserve representativeness of the complex sampling design [30]. Convergence diagnostics were evaluated using the fraction of missing information (FMI) and Monte Carlo (MC) error estimates obtained from the imputation variance table. Across all parameters, FMI values were approximately 0 and MC errors were close to 0, indicating negligible between-imputation variance and stable convergence across 20 datasets. Descriptive comparisons of observed and imputed data further confirmed that the imputed values closely reflected the distribution of the observed data, supporting the plausibility of the imputations. To assess robustness, complete-case sensitivity analyses were conducted using the original (non-imputed) datasets. The fully adjusted survey-weighted logistic regression models produced odds ratios nearly identical to the multiply imputed estimates—men: OR = 0.45 (95% CI: 0.23–0.88, p = 0.023) vs. MI OR = 0.45 (95% CI: 0.22–0.89, p = 0.025); women: OR = 0.43 (95% CI: 0.19–0.96, p = 0.041) vs. MI OR = 0.43 (95% CI: 0.19–0.97, p = 0.043). These consistent findings indicate that multiple imputation did not materially alter the effect estimates, confirming the robustness of the results. Multicollinearity was examined using variance inflation factors (VIF), with values > 5 indicating potential collinearity [31]. Statistical significance was set at p < 0.05 (two-tailed). Femoral-neck BMD served as the primary endpoint; analyses for other skeletal sites were considered exploratory. Results Descriptive characteristics of participants by dietary fat intake and sex A total of 612 men and 515 women who reported “usual” dietary intake were included in the analyses. The weighted prevalence of high-fat diet consumption demonstrated a significant sex difference, with men exhibiting higher rates (59.50%, 95% CI = 0.53–0.66, n = 323) compared to women (54%, 95% CI = 0.45–0.63, n = 262). Regarding bone health outcomes, the prevalence of osteopenia/osteoporosis was consistently higher among women than men across all anatomical sites, with the largest disparity observed at the femoral neck (men: 51.26%, 95% CI = 0.45–0.58, n = 308; women: 66.97%, 95% CI = 0.63–0.71, n = 328). Subsequent differences were seen in the total femur (men: 22.75%, 95% CI = 0.16–0.29, n = 158; women: 43.52%, 95% CI = 0.35–0.52, n = 219), trochanteric region (men: 22.56%, 95% CI = 0.16–0.29, n = 156; women: 39.10%, 95% CI = 0.32–0.46, n = 212), and intertrochanteric region (men: 19.35%, 95% CI = 0.13–0.26, n = 125; women: 38.01%, 95% CI = 0.30–0.46, n = 180). Across all measured skeletal sites, women showed a 1.7–2.0 fold higher prevalence of osteopenia or osteoporosis compared with men, with the femoral neck demonstrating the greatest absolute difference in prevalence (≈ 15%). These data indicate a higher baseline risk of low bone mass among women in this sample of older adults. Table 1 presents baseline characteristics according to dietary fat intake and sex. Among both sexes, participants with a high-fat diet exhibited higher total energy intake compared to those with a normal-fat diet (men: 2226 vs. 1898.50 kcal, p = 0.01; women: 1675.50 vs. 1505 kcal, p = 0.41), and increased BMI (men: 29.1 vs. 27.9 kg/m², p = 0.03; women: 28.9 vs. 28.1 kg/m², p = 0.08), with statistical significance observed only in men. They also had larger waist and hip circumferences (p < 0.05), along with significantly greater intake of saturated, monounsaturated, and polyunsaturated fatty acids (all p < 0.01). Abdominal obesity was markedly more prevalent in the high-fat diet group for both men (72.3% vs. 51.7%, p = 0.03) and women (83.7% vs. 64.7%, p = 0.01). Despite these differences, metabolic biomarkers—including LDL-C, total cholesterol, triglycerides, 25-OHD₃, total calcium, blood cadmium, and HS-CRP—remained statistically comparable between dietary groups (all p > 0.05). Sociodemographic characteristics, including age, marital status, smoking behavior, vigorous work activity, hypertension, diabetes, and family history of osteoporosis, were largely consistent across dietary patterns (all p > 0.05). Among women, those with a high-fat diet exhibited significantly higher alcohol consumption (91.22% vs. 82.93%, p = 0.02), whereas alcohol intake showed no significant difference in men (p = 0.71). Although a higher frequency of female hormone use was noted in the high-fat group (33.95% vs. 24.00%), this trend did not reach statistical significance (p = 0.13). Table 1Baseline characteristics of participants according to high-fat dietMenWomenNormal-Fat DietHigh-Fat DietP valueNormal-Fat DietHigh-Fat DietP valueAge (years)64 (11)65 (14)0.5363 (16)65 (16)0.53Race (%)0.010.35 Mexican American7.46 ± 2.622.77 ± 0.835.96 ± 1.533.36 ± 0.88 Other Hispanic10.67 ± 3.022.89 ± 0.688.03 ± 2.673.15 ± 1.53 Non-Hispanic White61.93 ± 6.5779.19 ± 2.9272.16 ± 3.5478.36 ± 5.45 Non-Hispanic Black8.78 ± 1.699.77 ± 1.875.42 ± 1.546.22 ± 2.03 Other Race11.15 ± 2.785.38 ± 1.608.42 ± 2.338.90 ± 4.09Married or living with partner (%)80.83 ± 3.2275.60 ± 2.820.2862.65 ± 5.4965.68 ± 6.220.77Smoking (%)49.65 ± 5.2254.52 ± 4.110.6635.05 ± 6.8237.99 ± 4.810.69Drinking (%)96.39 ± 1.1994.33 ± 2.410.7182.93 ± 2.8491.22 ± 2.340.02BMI (kg/m2)27.90 (6.10)29.10 (6.75)0.0328.10 (7.50)28.90 (7.90)0.08Waist circumference (cm)100.80 (14.70)106.10 (18.90)0.0298.20 (19.70)99 (15)0.10Hip circumference (cm)100.60 (12.10)104.20 (12.90)0.04104.40 (16.50)108.80 (14.70)0.04Total Femur BMD0.99 ± 0.020.99 ± 0.010.680.83 ± 0.020.86 ± 0.010.16Total Femur T-score (%)0.090.63≥ -176.28 ± 5.2577.60 ± 2.8650.90 ± 7.2761.47 ± 6.12-2.5 to -123.57 ± 5.2220.03 ± 2.7142.77 ± 7.1933.29 ± 5.81≤ -2.50.16 ± 0.162.37 ± 0.966.34 ± 1.605.24 ± 1.68Femoral Neck BMD0.77 (0.14)0.80 (0.19)0.270.69 (0.16)0.71 (0.18)0.23Femoral Neck T-score (%)0.070.21≥ -139.64 ± 5.2754.22 ± 3.5125.71 ± 5.0640.45 ± 3.48-2.5 to -156.58 ± 4.7140.14 ± 3.9862.89 ± 5.0950.89 ± 4.48≤ -2.53.78 ± 2.085.64 ± 2.1911.40 ± 1.748.66 ± 1.82Trochanter BMD0.74 (0.16)0.74 (0.17)0.940.63 (0.13)0.65 (0.17)0.10Trochanter T-score (%)0.770.38≥ -176.21 ± 4.6878.02 ± 3.5959.35 ± 5.2061.92 ± 4.03-2.5 to -122.94 ± 4.7620.87 ± 3.3838.34 ± 5.4133.11 ± 3.82≤ -2.50.85 ± 0.511.11 ± 0.582.31 ± 1.084.97 ± 1.78Intertrochanter BMD1.17 ± 0.021.18 ± 0.010.800.99 ± 0.021.03 ± 0.020.36Intertrochanter T-score (%)0.220.73≥ -179.83 ± 4.4481.03 ± 3.3858.85 ± 7.0964.91 ± 5.93-2.5 to -120.01 ± 4.4118.05 ± 3.2836.61 ± 6.1731.02 ± 5.75≤ -2.50.16 ± 0.160.91 ± 0.634.54 ± 1.854.07 ± 1.65Total Energy Intake (kcal)1898.50 (970)2226 (1014)0.011505 (545.50)1675.50 (727)0.41HDL-C (mg/dL)44 (18)46 (14)0.8055 (21)58 (21)0.31LDL-C (mg/dL)109.78 ± 3.96106.83 ± 4.090.64124.33 ± 4.10128.70 ± 4.740.52Total Cholesterol (mg/dL)179 (46)180 (65)0.79203 (57)209 (51)0.38Triglyceride (mg/dL)101 (85)100 (83)0.89104 (67)101 (66)0.87Total saturated fatty acids (gm)19.40 (12.71)31.87 (16.73)< 0.0115.11 (10.35)23.53 (14.08)< 0.01Total monounsaturated fatty acids (gm)20.35 (12.96)34.81 (19.48)< 0.0116.76 (9.94)25.48 (12.59)< 0.01Total polyunsaturated fatty acids (gm)14.10 (10.20)24.38 (13.75)< 0.0112.23 (5.22)17.93 (9.42)< 0.0125-OHD3 (nmol/L)65 (38.90)70.30 (36.70)0.8075.90 (48.60)80.50 (48.50)0.24Total Calcium (mmol/L)9.29 ± 0.039.31 ± 0.030.649.41 ± 0.049.40 ± 0.030.79Blood Cadmium (ug/L)0.29 (0.37)0.30 (0.37)0.920.405 (0.36)0.39 (0.31)0.997HS-CRP (mg/L)1.75 (3.23)2 (3.79)0.311.97 (3.52)2.17 (2.79)0.84Abdominal Obesity (%)51.69 ± 6.3272.31 ± 4.270.0364.70 ± 4.8483.65 ± 3.530.01Hypertension (%)60.23 ± 3.6064.18 ± 6.500.4732.40 ± 4.3427.99 ± 2.770.35Diabetes (%)30.74 ± 5.7026.22 ± 4.500.7018.48 ± 4.1418.73 ± 2.450.78Family history of osteoporosis (%)8.74 ± 2.2715.58 ± 3.860.1923.82 ± 4.3618.90 ± 4.940.79Vigorous work activity (%)23.97 ± 5.1236.99 ± 6.410.2012.78 ± 3.6318.53 ± 5.660.31Female hormones use (%)———24 ± 4.1133.95 ± 5.170.13weighted proportions ± standard error (SE) for categorical variables and as weighted means ± SE or median (interquartile range) for continuous variables. BMI: body mass index; BMD: bone mineral density; HDL-C: high density lipoprotein-cholesterol; LDL-C: low density lipoprotein-cholesterol; 25-OHD3: 25-hydroxyvitamin D3; HS-CRP: High-Sensitivity C-Reactive Protein Table 1Baseline characteristics of participants according to high-fat dietMenWomenNormal-Fat DietHigh-Fat DietP valueNormal-Fat DietHigh-Fat DietP valueAge (years)64 (11)65 (14)0.5363 (16)65 (16)0.53Race (%)0.010.35 Mexican American7.46 ± 2.622.77 ± 0.835.96 ± 1.533.36 ± 0.88 Other Hispanic10.67 ± 3.022.89 ± 0.688.03 ± 2.673.15 ± 1.53 Non-Hispanic White61.93 ± 6.5779.19 ± 2.9272.16 ± 3.5478.36 ± 5.45 Non-Hispanic Black8.78 ± 1.699.77 ± 1.875.42 ± 1.546.22 ± 2.03 Other Race11.15 ± 2.785.38 ± 1.608.42 ± 2.338.90 ± 4.09Married or living with partner (%)80.83 ± 3.2275.60 ± 2.820.2862.65 ± 5.4965.68 ± 6.220.77Smoking (%)49.65 ± 5.2254.52 ± 4.110.6635.05 ± 6.8237.99 ± 4.810.69Drinking (%)96.39 ± 1.1994.33 ± 2.410.7182.93 ± 2.8491.22 ± 2.340.02BMI (kg/m2)27.90 (6.10)29.10 (6.75)0.0328.10 (7.50)28.90 (7.90)0.08Waist circumference (cm)100.80 (14.70)106.10 (18.90)0.0298.20 (19.70)99 (15)0.10Hip circumference (cm)100.60 (12.10)104.20 (12.90)0.04104.40 (16.50)108.80 (14.70)0.04Total Femur BMD0.99 ± 0.020.99 ± 0.010.680.83 ± 0.020.86 ± 0.010.16Total Femur T-score (%)0.090.63≥ -176.28 ± 5.2577.60 ± 2.8650.90 ± 7.2761.47 ± 6.12-2.5 to -123.57 ± 5.2220.03 ± 2.7142.77 ± 7.1933.29 ± 5.81≤ -2.50.16 ± 0.162.37 ± 0.966.34 ± 1.605.24 ± 1.68Femoral Neck BMD0.77 (0.14)0.80 (0.19)0.270.69 (0.16)0.71 (0.18)0.23Femoral Neck T-score (%)0.070.21≥ -139.64 ± 5.2754.22 ± 3.5125.71 ± 5.0640.45 ± 3.48-2.5 to -156.58 ± 4.7140.14 ± 3.9862.89 ± 5.0950.89 ± 4.48≤ -2.53.78 ± 2.085.64 ± 2.1911.40 ± 1.748.66 ± 1.82Trochanter BMD0.74 (0.16)0.74 (0.17)0.940.63 (0.13)0.65 (0.17)0.10Trochanter T-score (%)0.770.38≥ -176.21 ± 4.6878.02 ± 3.5959.35 ± 5.2061.92 ± 4.03-2.5 to -122.94 ± 4.7620.87 ± 3.3838.34 ± 5.4133.11 ± 3.82≤ -2.50.85 ± 0.511.11 ± 0.582.31 ± 1.084.97 ± 1.78Intertrochanter BMD1.17 ± 0.021.18 ± 0.010.800.99 ± 0.021.03 ± 0.020.36Intertrochanter T-score (%)0.220.73≥ -179.83 ± 4.4481.03 ± 3.3858.85 ± 7.0964.91 ± 5.93-2.5 to -120.01 ± 4.4118.05 ± 3.2836.61 ± 6.1731.02 ± 5.75≤ -2.50.16 ± 0.160.91 ± 0.634.54 ± 1.854.07 ± 1.65Total Energy Intake (kcal)1898.50 (970)2226 (1014)0.011505 (545.50)1675.50 (727)0.41HDL-C (mg/dL)44 (18)46 (14)0.8055 (21)58 (21)0.31LDL-C (mg/dL)109.78 ± 3.96106.83 ± 4.090.64124.33 ± 4.10128.70 ± 4.740.52Total Cholesterol (mg/dL)179 (46)180 (65)0.79203 (57)209 (51)0.38Triglyceride (mg/dL)101 (85)100 (83)0.89104 (67)101 (66)0.87Total saturated fatty acids (gm)19.40 (12.71)31.87 (16.73)< 0.0115.11 (10.35)23.53 (14.08)< 0.01Total monounsaturated fatty acids (gm)20.35 (12.96)34.81 (19.48)< 0.0116.76 (9.94)25.48 (12.59)< 0.01Total polyunsaturated fatty acids (gm)14.10 (10.20)24.38 (13.75)< 0.0112.23 (5.22)17.93 (9.42)< 0.0125-OHD3 (nmol/L)65 (38.90)70.30 (36.70)0.8075.90 (48.60)80.50 (48.50)0.24Total Calcium (mmol/L)9.29 ± 0.039.31 ± 0.030.649.41 ± 0.049.40 ± 0.030.79Blood Cadmium (ug/L)0.29 (0.37)0.30 (0.37)0.920.405 (0.36)0.39 (0.31)0.997HS-CRP (mg/L)1.75 (3.23)2 (3.79)0.311.97 (3.52)2.17 (2.79)0.84Abdominal Obesity (%)51.69 ± 6.3272.31 ± 4.270.0364.70 ± 4.8483.65 ± 3.530.01Hypertension (%)60.23 ± 3.6064.18 ± 6.500.4732.40 ± 4.3427.99 ± 2.770.35Diabetes (%)30.74 ± 5.7026.22 ± 4.500.7018.48 ± 4.1418.73 ± 2.450.78Family history of osteoporosis (%)8.74 ± 2.2715.58 ± 3.860.1923.82 ± 4.3618.90 ± 4.940.79Vigorous work activity (%)23.97 ± 5.1236.99 ± 6.410.2012.78 ± 3.6318.53 ± 5.660.31Female hormones use (%)———24 ± 4.1133.95 ± 5.170.13weighted proportions ± standard error (SE) for categorical variables and as weighted means ± SE or median (interquartile range) for continuous variables. BMI: body mass index; BMD: bone mineral density; HDL-C: high density lipoprotein-cholesterol; LDL-C: low density lipoprotein-cholesterol; 25-OHD3: 25-hydroxyvitamin D3; HS-CRP: High-Sensitivity C-Reactive Protein Baseline characteristics of participants according to high-fat diet weighted proportions ± standard error (SE) for categorical variables and as weighted means ± SE or median (interquartile range) for continuous variables. BMI: body mass index; BMD: bone mineral density; HDL-C: high density lipoprotein-cholesterol; LDL-C: low density lipoprotein-cholesterol; 25-OHD3: 25-hydroxyvitamin D3; HS-CRP: High-Sensitivity C-Reactive Protein Sex-stratified analyses further revealed modest, site-specific differences in bone health. Among women, those with a high-fat diet exhibited slightly higher BMD at the femoral neck (0.71 vs. 0.69 g/cm²), total femur (0.86 vs. 0.83 g/cm²), trochanter (0.65 vs. 0.63 g/cm²), and intertrochanteric region (1.03 vs. 0.99 g/cm²), accompanied by a generally lower prevalence of osteopenia/osteoporosis, except at the trochanteric site, where osteoporosis was more frequent in the high-fat group (4.97% vs. 2.31%). None of these differences reached statistical significance (all P > 0.05). In men, no meaningful associations were found between high-fat diet and bone health. BMD values were nearly identical across sites—total femur (0.99 vs. 0.99 g/cm²), femoral neck (0.80 vs. 0.77 g/cm²), trochanter (0.74 vs. 0.74 g/cm²), and intertrochanteric region (1.18 vs. 1.17 g/cm²)—and the prevalence of osteopenia/osteoporosis did not differ significantly (all P > 0.05). Association of High-Fat diet with Osteopenia/Osteoporosis and BMD In multivariable survey-weighted logistic regression analyses (Table 2), a high-fat diet (> 35%E) was associated with significantly lower odds of femoral neck osteopenia/osteoporosis in both men (OR = 0.45, 95% CI = 0.22–0.89, p = 0.03) and women (OR = 0.43, 95% CI = 0.19–0.97, p = 0.04) after full adjustment (Model III), indicating a potential protective effect of higher dietary fat intake on femoral neck bone health. No significant associations were observed at the total femur, trochanter, or intertrochanteric regions (all p > 0.05), suggesting that this effect may be site-specific and most pronounced at the femoral neck. These null findings may reflect site-specific metabolic or mechanical differences between cortical and trabecular bone or limited statistical power in less variable skeletal regions. Table 2Association between dietary fat intake and femoral neck osteopenia/osteoporosisHigh-fat diet(> 35%E vs. ≤ 35%E)Model IModel IIModel IIIOR (95% CI)p-valueOR (95% CI)p-valueOR (95% CI)p-valueMenFemoral neck0.57 (0.32–1.02)0.060.49 (0.25–0.94)0.030.45 (0.22–0.89)0.03Total femur0.88 (0.49–1.58)0.660.76 (0.39–1.49)0.400.70 (0.36–1.33)0.25Femoral trochanter0.86 (0.46–1.63)0.620.79 (0.38–1.64)0.500.71 (0.38–1.34)0.27Femoral intertrochanteric0.88 (0.47–1.64)0.660.73 (0.37–1.43)0.330.65 (0.33–1.28)0.19WomenFemoral neck0.52 (0.25–1.10)0.080.50 (0.21–1.21)0.120.43 (0.19–0.97)0.04Total femur0.71 (0.30–1.72)0.420.67 (0.25–1.83)0.410.64 (0.27–1.51)0.29Femoral trochanter0.83 (0.48–1.43)0.480.95 (0.49–1.84)0.871.01 (0.53–1.91)0.97Femoral intertrochanteric0.86 (0.34–2.18)0.740.80 (0.28–2.83)0.670.77 (0.28–2.06)0.57OR = odds ratio; CI = confidence interval. Model I: unadjusted. Model II: partially adjusted for age, race, total energy intake, smoking, alcohol consumption, vigorous activity, and estrogen replacement therapy. Model III: fully adjusted for Model II covariates + marital status, HDL-C, LDL-C, total cholesterol, triglycerides, 25-OHD₃, total calcium, HS-CRP, blood cadmium, hypertension, diabetes, and family history of osteoporosis. Femoral neck osteopenia/osteoporosis was the primary endpoint Table 2Association between dietary fat intake and femoral neck osteopenia/osteoporosisHigh-fat diet(> 35%E vs. ≤ 35%E)Model IModel IIModel IIIOR (95% CI)p-valueOR (95% CI)p-valueOR (95% CI)p-valueMenFemoral neck0.57 (0.32–1.02)0.060.49 (0.25–0.94)0.030.45 (0.22–0.89)0.03Total femur0.88 (0.49–1.58)0.660.76 (0.39–1.49)0.400.70 (0.36–1.33)0.25Femoral trochanter0.86 (0.46–1.63)0.620.79 (0.38–1.64)0.500.71 (0.38–1.34)0.27Femoral intertrochanteric0.88 (0.47–1.64)0.660.73 (0.37–1.43)0.330.65 (0.33–1.28)0.19WomenFemoral neck0.52 (0.25–1.10)0.080.50 (0.21–1.21)0.120.43 (0.19–0.97)0.04Total femur0.71 (0.30–1.72)0.420.67 (0.25–1.83)0.410.64 (0.27–1.51)0.29Femoral trochanter0.83 (0.48–1.43)0.480.95 (0.49–1.84)0.871.01 (0.53–1.91)0.97Femoral intertrochanteric0.86 (0.34–2.18)0.740.80 (0.28–2.83)0.670.77 (0.28–2.06)0.57OR = odds ratio; CI = confidence interval. Model I: unadjusted. Model II: partially adjusted for age, race, total energy intake, smoking, alcohol consumption, vigorous activity, and estrogen replacement therapy. Model III: fully adjusted for Model II covariates + marital status, HDL-C, LDL-C, total cholesterol, triglycerides, 25-OHD₃, total calcium, HS-CRP, blood cadmium, hypertension, diabetes, and family history of osteoporosis. Femoral neck osteopenia/osteoporosis was the primary endpoint Association between dietary fat intake and femoral neck osteopenia/osteoporosis OR = odds ratio; CI = confidence interval. Model I: unadjusted. Model II: partially adjusted for age, race, total energy intake, smoking, alcohol consumption, vigorous activity, and estrogen replacement therapy. Model III: fully adjusted for Model II covariates + marital status, HDL-C, LDL-C, total cholesterol, triglycerides, 25-OHD₃, total calcium, HS-CRP, blood cadmium, hypertension, diabetes, and family history of osteoporosis. Femoral neck osteopenia/osteoporosis was the primary endpoint When total fat intake (%E) was modeled as a continuous variable (Table 3), each 1%E increase in total fat intake was associated with lower odds of femoral neck osteopenia/osteoporosis in both men (OR = 0.96, 95% CI = 0.92–0.99, p = 0.03) and women (OR = 0.95, 95% CI = 0.90–0.99, p = 0.03), supporting a dose–response like trend between dietary fat intake and bone integrity. Restricted cubic spline analyses indicated no evidence of nonlinearity among men (P = 0.44), with a steady decline in risk as the percentage of energy from dietary fat increased (Fig. 2A). In contrast, a significant nonlinear association was observed in women (P < 0.01), characterized by a decreasing risk up to approximately 40% of energy from dietary fat, beyond which the protective trend plateaued or slightly reversed at higher intake levels (Fig. 2B). Table 3Sensitivity analyses of dietary fat intake and femoral neck osteopenia/osteoporosisAnalysisVariableEffect estimate (95% CI)p-valueContinuousdose–responseTotal fat intake (%E, per 1%E increase), menOR = 0.96 (0.92–0.99)0.03Total fat intake (%E, per 1%E increase), womenOR = 0.95 (0.90–0.99)0.03RCS nonlinearity (global F-test), menF = 0.840.44RCS nonlinearity (global F-test), womenF = 7.23< 0.01Linear validationHigh-fat diet (> 35%E vs. ≤35%E), menβ = 0 0.020.07High-fat diet (> 35%E vs. ≤35%E), womenβ = 0.030.05Total fat intake (%E, per 1%E increase), menβ = 0.0010.19Total fat intake (%E, per 1%E increase), womenβ = 0.0020.09Fat subtype analyses (%E)Saturated fatty acids (SFA), menOR = 0.98 (0.95–1.02)0.34Saturated fatty acids (SFA), womenOR = 0.92 (0.87–0.98)< 0.01Monounsaturated fatty acids (MUFA), menOR = 0.99 (0.95–1.02)0.43Monounsaturated fatty acids (MUFA), womenOR = 0.98 (0.93–1.02)0.26Polyunsaturated fatty acids (PUFA), menOR = 0.96 (0.92–1.01)0.10Polyunsaturated fatty acids (PUFA), womenOR = 0.98 (0.94–1.02)0.28Isocaloric substitution(5%E replacement)Fat replacing carbohydrate, menOR = 0.79 (0.64–0.97)0.03Fat replacing protein, menOR = 0.78 (0.60–1.01)0.06Fat replacing carbohydrate, womenOR = 0.76 (0.60–0.97)0.03Fat replacing protein, womenOR = 0.76 (0.55–1.07)0.10High-fat diet subgroup analysis35% < total fat ≤ 40%E, menOR = 0.35 (0.16–0.76)0.0140% < total fat ≤ 45%E, menOR = 0.64 (0.28–1.47)0.27> 45%E, menOR = 0.45 (0.17–1.19)0.1035% < total fat ≤ 40%E, womenOR = 0.47 (0.21–1.05)0.0640% < total fat ≤ 45%E, womenOR = 0.22 (0.06–0.77)0.02> 45%E, womenOR = 0.77 (0.32–1.84)0.53Mediation(body composition)BMI-adjusted model, menOR = 0.54 (0.29–1.00)0.05BMI-adjusted model, womenOR = 0.39 (0.17–0.92)0.03Including non-usual intakeHigh-fat diet (> 35%E vs. ≤35%E), menOR = 0.62 (0.32–1.18)0.13High-fat diet (> 35%E vs. ≤35%E), womenOR = 0.77 (0.35–1.67)0.48All analyses were conducted using survey-weighted logistic regression for femoral neck osteopenia/osteoporosis and linear regression for continuous BMD, with full adjustment (Model III) for age, race, total energy intake, smoking, alcohol, vigorous activity, estrogen replacement therapy, marital status, HDL-C, LDL-C, total cholesterol, triglycerides, 25-OHD₃, total calcium, HS-CRP, blood cadmium, hypertension, diabetes, and family history of osteoporosis. Femoral neck osteopenia/osteoporosis was the primary endpoint; other tests are exploratory or sensitivity analyses. Reference group for high-fat subgroup analysis: total fat ≤ 35%E. SFA, MUFA, and PUFA were modeled per 1%E increase in energy contribution Table 3Sensitivity analyses of dietary fat intake and femoral neck osteopenia/osteoporosisAnalysisVariableEffect estimate (95% CI)p-valueContinuousdose–responseTotal fat intake (%E, per 1%E increase), menOR = 0.96 (0.92–0.99)0.03Total fat intake (%E, per 1%E increase), womenOR = 0.95 (0.90–0.99)0.03RCS nonlinearity (global F-test), menF = 0.840.44RCS nonlinearity (global F-test), womenF = 7.23< 0.01Linear validationHigh-fat diet (> 35%E vs. ≤35%E), menβ = 0 0.020.07High-fat diet (> 35%E vs. ≤35%E), womenβ = 0.030.05Total fat intake (%E, per 1%E increase), menβ = 0.0010.19Total fat intake (%E, per 1%E increase), womenβ = 0.0020.09Fat subtype analyses (%E)Saturated fatty acids (SFA), menOR = 0.98 (0.95–1.02)0.34Saturated fatty acids (SFA), womenOR = 0.92 (0.87–0.98)< 0.01Monounsaturated fatty acids (MUFA), menOR = 0.99 (0.95–1.02)0.43Monounsaturated fatty acids (MUFA), womenOR = 0.98 (0.93–1.02)0.26Polyunsaturated fatty acids (PUFA), menOR = 0.96 (0.92–1.01)0.10Polyunsaturated fatty acids (PUFA), womenOR = 0.98 (0.94–1.02)0.28Isocaloric substitution(5%E replacement)Fat replacing carbohydrate, menOR = 0.79 (0.64–0.97)0.03Fat replacing protein, menOR = 0.78 (0.60–1.01)0.06Fat replacing carbohydrate, womenOR = 0.76 (0.60–0.97)0.03Fat replacing protein, womenOR = 0.76 (0.55–1.07)0.10High-fat diet subgroup analysis35% < total fat ≤ 40%E, menOR = 0.35 (0.16–0.76)0.0140% < total fat ≤ 45%E, menOR = 0.64 (0.28–1.47)0.27> 45%E, menOR = 0.45 (0.17–1.19)0.1035% < total fat ≤ 40%E, womenOR = 0.47 (0.21–1.05)0.0640% < total fat ≤ 45%E, womenOR = 0.22 (0.06–0.77)0.02> 45%E, womenOR = 0.77 (0.32–1.84)0.53Mediation(body composition)BMI-adjusted model, menOR = 0.54 (0.29–1.00)0.05BMI-adjusted model, womenOR = 0.39 (0.17–0.92)0.03Including non-usual intakeHigh-fat diet (> 35%E vs. ≤35%E), menOR = 0.62 (0.32–1.18)0.13High-fat diet (> 35%E vs. ≤35%E), womenOR = 0.77 (0.35–1.67)0.48All analyses were conducted using survey-weighted logistic regression for femoral neck osteopenia/osteoporosis and linear regression for continuous BMD, with full adjustment (Model III) for age, race, total energy intake, smoking, alcohol, vigorous activity, estrogen replacement therapy, marital status, HDL-C, LDL-C, total cholesterol, triglycerides, 25-OHD₃, total calcium, HS-CRP, blood cadmium, hypertension, diabetes, and family history of osteoporosis. Femoral neck osteopenia/osteoporosis was the primary endpoint; other tests are exploratory or sensitivity analyses. Reference group for high-fat subgroup analysis: total fat ≤ 35%E. SFA, MUFA, and PUFA were modeled per 1%E increase in energy contribution Sensitivity analyses of dietary fat intake and femoral neck osteopenia/osteoporosis Continuous dose–response Isocaloric substitution (5%E replacement) Mediation (body composition) All analyses were conducted using survey-weighted logistic regression for femoral neck osteopenia/osteoporosis and linear regression for continuous BMD, with full adjustment (Model III) for age, race, total energy intake, smoking, alcohol, vigorous activity, estrogen replacement therapy, marital status, HDL-C, LDL-C, total cholesterol, triglycerides, 25-OHD₃, total calcium, HS-CRP, blood cadmium, hypertension, diabetes, and family history of osteoporosis. Femoral neck osteopenia/osteoporosis was the primary endpoint; other tests are exploratory or sensitivity analyses. Reference group for high-fat subgroup analysis: total fat ≤ 35%E. SFA, MUFA, and PUFA were modeled per 1%E increase in energy contribution Fig. 2Restricted cubic spline analysis of the association between total dietary fat intake and femoral neck osteopenia/osteoporosis in men (A) and women (B). The solid line represents adjusted ORs and the shaded area indicates the 95% confidence interval. The vertical dashed line at 35%E denotes the reference level of total fat intake Restricted cubic spline analysis of the association between total dietary fat intake and femoral neck osteopenia/osteoporosis in men (A) and women (B). The solid line represents adjusted ORs and the shaded area indicates the 95% confidence interval. The vertical dashed line at 35%E denotes the reference level of total fat intake Complementary linear regression analyses using continuous BMD values showed directionally consistent associations: a high-fat diet was positively related to femoral neck BMD in men (β = 0.02, p = 0.07) and women (β = 0.03, p = 0.05), although these associations were of borderline statistical significance (Table 3). When total fat intake (%E) was analyzed as a continuous variable, each 1%E increase was associated with a small positive change in femoral neck BMD (men: β = 0.001, p = 0.19; women: β = 0.002, p = 0.09), further indicating a positive trend in femoral neck BMD with increasing fat consumption. Overall, these findings indicate that the association between higher fat intake and femoral neck bone density is modest in magnitude but consistent across multiple analytical approaches, underscoring the robustness of the observed relationship. Sensitivity and exploratory analyses (Table 3) Mediation analyses incorporating body composition variables (BMI, waist and hip circumference, and abdominal obesity) did not materially change the associations between high-fat diet and bone outcomes (men: OR = 0.54, p = 0.05; women: OR = 0.39, p = 0.03), suggesting that the observed associations were largely independent of adiposity-related factors. In fatty acid subtype analyses, substituting saturated, monounsaturated, or polyunsaturated fats for total fat yielded comparable results in men (all p > 0.05), whereas among women, higher saturated fat intake was significantly associated with lower odds of femoral neck osteopenia/osteoporosis (OR = 0.92, 95% CI = 0.87–0.98, p < 0.01), while MUFA and PUFA were not significant (p > 0.05). These results suggest that saturated fat may contribute more prominently to bone preservation among women. In isocaloric substitution models, replacing 5% of total energy from carbohydrates with total fat was associated with reduced odds of femoral neck osteopenia/osteoporosis in both sexes (men: OR = 0.79, 95% CI = 0.64–0.97, p = 0.03; women: OR = 0.76, 95% CI = 0.60–0.97, p = 0.03), whereas substitution of protein showed borderline significance (men: p = 0.06; women: p = 0.10). These findings suggest that partial replacement of carbohydrates with dietary fat may confer a modest protective association on bone integrity. Exploratory dose–response analyses using categorical fat intake levels (reference ≤ 35%E) revealed distinct sex-specific patterns. Among men, the lowest odds of femoral neck osteopenia/osteoporosis were observed at a moderately high fat intake of 35–40%E (OR = 0.35, p = 0.01, n = 151), while in women, the strongest protective association occurred at a slightly higher intake range of 40–45%E (OR = 0.22, p = 0.02, n = 89). At higher fat intake levels (> 45%E; women = 47, men = 63), the associations plateaued and lost statistical significance (p > 0.10). This pattern, consistent with the spline-based results, indicates a dose-dependent yet nonlinear association, suggesting that a moderate-to-high fat intake range may represent an optimal threshold for maintaining femoral neck bone health. Including participants with non-usual dietary intake did not materially alter the findings (men: OR = 0.62, p = 0.13, n = 966; women: OR = 0.77, p = 0.48, n = 899), confirming the robustness of the observed associations. Discussion This study provides population-based evidence that dietary fat intake is associated with BMD and osteopenia/osteoporosis risk in a sex- and site-specific manner among U.S. adults aged ≥ 50 years. Using nationally representative NHANES data, we found that a high-fat diet (> 35%E) was associated with lower odds of femoral neck osteopenia/osteoporosis in both men and women after multivariable adjustment, with no statistically significant adverse associations observed at other skeletal sites, although confidence intervals were wide. Our results align with previous population-based studies showing that moderate dietary fat intake exerts neutral or beneficial effects on bone health, likely through mechanisms involving improved energy metabolism and enhanced calcium and vitamin D absorption [13, 15, 16, 32]. Collectively, these findings indicate a consistent, though modest, protective association between dietary fat intake and femoral neck bone health in older adults. This site-specific effect may reflect the femoral neck’s higher trabecular bone content and metabolic activity compared with other femoral regions, making it more responsive to dietary and hormonal influences [33, 34]. Moreover, partial replacement of carbohydrates with dietary fat appeared to confer a modest protective effect on bone integrity. Overall, these findings highlight that the relationship between dietary fat and bone health is context-dependent—shaped by age, hormonal status, skeletal site, and overall dietary composition—rather than uniformly positive or negative. Although moderate fat intake appeared to be associated with better skeletal preservation in both sexes, the observed dose–response patterns differed between men and women. Women had a higher baseline prevalence of osteopenia/osteoporosis—approximately 1.7–2.0 times that of men across femoral sites—and exhibited a suggestive nonlinear trend, with the lowest estimated risk at moderate fat intake levels (around 40–45%E) and a possible plateau at higher intakes. In contrast, men showed an approximately linear inverse association, with lower risk observed across increasing fat intake up to about 35–40%E. These findings may indicate potential sex-related variation in skeletal sensitivity to dietary fat; however, given the limited sample sizes within subgroups and associated statistical uncertainty, these results should be interpreted as exploratory and hypothesis-generating. The apparent sex difference in dose–response patterns could reflect underlying hormonal or metabolic factors rather than direct causal effects. Prior research suggests that dietary fats may interact with estrogen-mediated bone turnover pathways, including Wnt/β-catenin and RANKL/OPG signaling [33, 35, 36]. Lipid intake might also influence estrogen bioavailability through adipose aromatization and steroid hormone transport [33, 37], processes that could be particularly relevant in postmenopausal women experiencing hormonal decline [33, 34, 38]. Among men, the relatively weaker associations observed might be attributed to the predominant influence of mechanical loading and androgen-mediated regulation [34, 39]; however, these mechanisms were not directly examined in this study. Marrow adipose tissue (MAT) may also contribute to sex- and site-specific differences in skeletal metabolism. Previous studies have shown that MAT tends to increase with age and may influence bone remodeling through inflammatory and energy-regulatory pathways [40, 41]. In estrogen-deficient states, such as postmenopause, MAT could potentially serve as a local energy reservoir at metabolically active sites like the femoral neck. These explanations remain speculative and were not directly tested in our analysis. Future longitudinal and mechanistic studies are warranted to confirm whether the observed associations represent causal or compensatory biological processes. From a public health perspective, these findings suggest that maintaining overall macronutrient balance—rather than pursuing strict fat restriction—may be relevant for bone health management in older adults. For postmenopausal women in particular, ensuring adequate but not excessive fat intake could represent a practical and physiologically reasonable component of osteoporosis prevention. Clinicians and policymakers should therefore promote balanced dietary patterns that include adequate fat intake from primarily healthy sources (e.g., monounsaturated and polyunsaturated fats), while acknowledging that moderate levels of saturated fat intake may not be detrimental to bone health. Such balanced nutritional approaches could help preserve bone integrity and reduce the burden of osteoporosis in aging populations. Nevertheless, these implications should be regarded as exploratory rather than prescriptive. Given the cross-sectional design of this study, causal inference cannot be established, and residual confounding may remain. Future longitudinal and mechanistic studies are needed to confirm these associations and to clarify the biological pathways through which dietary fat may influence bone health. Limitations Several limitations should be acknowledged. First, causal inference cannot be established due to the cross-sectional design. Although a protective association between dietary fat intake and femoral neck osteopenia/osteoporosis was observed, longitudinal and mechanistic studies are needed to clarify causality. Second, dietary data were derived from 24-hour recalls, which are subject to recall bias and day-to-day variability, particularly among older adults. Third, DXA assessment captures bone mineral density but not bone microarchitecture or quality; future studies incorporating HR-pQCT could provide deeper insights into structural adaptations. Fourth, relatively small subgroup sizes at extreme fat intake levels may have limited statistical power, underscoring the need for larger, well-powered studies or pooled analyses. Lastly, site-specific discrepancies—protective effects at the femoral neck but not at other regions—suggest heterogeneous skeletal responses, likely reflecting regional variation in trabecular composition, metabolic activity, and mechanical loading. Collectively, these limitations underscore the need for longitudinal and mechanistic research to confirm causality and elucidate the biological pathways underlying the observed sex- and site-specific associations. Conclusions In conclusion, this cross-sectional analysis of a nationally representative sample suggests that dietary fat intake may be modestly associated with bone mineral density and osteopenia/osteoporosis risk in a sex- and site-specific manner among U.S. adults aged ≥ 50 years. A higher proportion of dietary fat intake was linked to a lower likelihood of femoral neck osteopenia/osteoporosis in both men and women, while no adverse associations were observed at other skeletal sites. Exploratory dose–response analyses indicated that moderate fat intake appeared most favorable for femoral bone health, showing distinct sex-specific trends. These findings should be interpreted cautiously given the cross-sectional design and limited subgroup power. They point to a potential beneficial role of balanced fat intake in maintaining skeletal health among older adults, warranting confirmation in future longitudinal and mechanistic studies.