Association between lipid accumulation product and visceral adiposity index and rheumatoid arthritis: NHANES 1999–2018 Authors: Jinye Xu, Yunsheng Xu, Fuxin Wei Published: 2026 ABSTRACT AbstractBackgroundWhile the Lipid Accumulation Product (LAP) and Visceral Adiposity Index (VAI) offer improved assessment of visceral fat distribution compared to traditional measures, their connection with rheumatoid arthritis (RA) has not been thoroughly explored. This research aimed to assess the links between LAP, VAI, and RA occurrence using nationally representative datasets.MethodsThis cross-sectional study analyzed information from the National Health and Nutrition Examination Survey (NHANES) covering the years 1999–2018. RA status was determined based on self-reported physician diagnosis. Sex-specific formulas were employed to calculate LAP and VAI. Statistical techniques included weighted multivariable logistic regression with three progressive models, restricted cubic spline (RCS) analysis, piecewise linear regression, and subgroup analyses.ResultsThe study comprised 15,918 individuals, including 1,988 RA cases and 13,930 non-RA controls. RA patients demonstrated distinct demographic and clinical profiles compared to non-RA participants, exhibiting older age, greater female representation, reduced educational attainment (P = 0.03), and elevated frequencies of smoking, hypertension, and diabetes (all P < 0.001). Both LAP and VAI measurements were markedly elevated in the RA group (P < 0.001). After comprehensive covariate adjustment, LAP maintained a significant association with RA prevalence (OR = 1.002, P < 0.001). Individuals in the highest LAP tertile displayed 41.0% greater RA prevalence (OR = 1.410, P < 0.001), demonstrating a clear dose-response pattern (P for trend < 0.001). RCS analysis identified a significant nonlinear LAP-RA relationship (P for nonlinearity < 0.001), with a threshold effect at LAP = 43.45. Below this value, the association was notably stronger (OR = 1.0028, P < 0.001), whereas above it the relationship plateaued (P = 0.096). In contrast, VAI exhibited no significant association with RA (P for overall = 0.397) or nonlinear pattern (P for nonlinearity = 0.864). Stratified analyses revealed that hypertension status significantly modified the LAP-RA association (P for interaction = 0.040), with hypertensive individuals showing more pronounced effects. Borderline age-related differences were also noted, with stronger associations among younger participants (< 60 years) and in non-Hispanic White and non-Hispanic Black subgroups.ConclusionThis research indicates a potential link between LAP and RA occurrence, possibly highlighting the influence of abdominal fat deposition in RA development. Being an easily obtainable indicator that solely necessitates waist measurement and triglyceride assessment, LAP could function as an efficient preliminary screening method for identifying RA susceptibility in general healthcare environments. Additionally, it might contribute valuable insights for developing preventive approaches targeting metabolic factors in RA management.Clinical trial numberNot applicable.Keywords: Lipid accumulation product, Visceral adiposity index, Rheumatoid arthritis, National Health and Nutrition Examination Survey, Cross-sectional studies Background Rheumatoid arthritis (RA) represents a persistent autoimmune disorder marked by inflammation, impacting roughly 17.6 million individuals globally, with standardized prevalence rates reaching 208.8 cases per 100,000 people [1]. This multisystem condition shows a female predominance and manifests through deteriorating joint damage, functional impairment, and elevated mortality rates [2]. Although genetic predisposition and environmental triggers influence disease development, recent studies emphasize the important role of metabolic disturbances, especially excessive adiposity, as key regulators of immune function and clinical outcomes [3]. Functioning as a dynamic endocrine system, fat deposits release inflammatory mediators including tumor necrosis factor-α (TNF-α), interleukin-6 (IL-6), and monocyte chemoattractant protein-1 (MCP-1) that fuel chronic inflammation central to RA pathology [4, 5]. Such pro-inflammatory conditions facilitate autoimmune reactions and articular deterioration, demonstrating the pathogenic interplay between metabolic abnormalities and rheumatic disorders [6]. Among fat depots, intra-abdominal visceral adipose tissue demonstrates heightened inflammatory activity relative to subcutaneous fat, correlating more significantly with metabolic derangements and cardiovascular risks [7]. Conventional obesity indicators such as body mass index (BMI) and waist measurements exhibit fundamental shortcomings in differentiating subcutaneous from intra-abdominal fat deposits [8]. These standard parameters inadequately represent the diverse patterns of fat distribution and metabolic function, especially concerning visceral fat’s role as a key source of pro-inflammatory mediators [9]. Contemporary composite markers combining physical measurements with metabolic data have surfaced as improved alternatives for evaluating obesity-related health hazards. The Lipid Accumulation Product (LAP), derived from waist circumference (WC) and triglyceride (TG) levels, offers enhanced precision in quantifying visceral fat buildup and related metabolic abnormalities [10]. Similarly, the Visceral Adiposity Index (VAI), a sex-specific formula incorporating BMI, waist size, and blood lipid parameters, more accurately represents the physiological impact of fat distribution patterns than conventional measures [11]. These sophisticated metrics show improved predictive value for metabolic disorders and cardiovascular conditions relative to traditional approaches. Experimental studies indicate that visceral fat accumulation stimulates Th1 and Th17 cell activation while inhibiting regulatory T cell activity, immunological mechanisms fundamentally involved in RA development [12]. While these pathophysiological connections are established, clinical research exploring links between modern adiposity markers and RA remains scarce, with most investigations relying on basic physical measurements that produce contradictory results. It was hypothesized that LAP and VAI, as refined markers of visceral obesity, would show positive correlations with RA occurrence, with LAP potentially exhibiting stronger associations due to its closer reflection of lipid storage abnormalities and abnormal fat distribution. Currently, no research has comprehensively assessed both LAP and VAI in relation to RA or investigated possible nonlinear correlations and critical thresholds within a nationally representative sample. Consequently, this investigation sought to analyze connections between LAP, VAI, and RA prevalence, identify potential nonlinear response patterns, and determine demographic segments where these relationships might be most evident, utilizing NHANES datasets spanning 1999–2018. Methods Study design and participants This investigation employed a cross-sectional analytical methodology, leveraging publicly available datasets from the National Health and Nutrition Examination Survey (NHANES) spanning 1999–2018. Administered by the National Center for Health Statistics (NCHS) under the Centers for Disease Control and Prevention (CDC), NHANES employs an advanced, stratified, multistage probability sampling framework to generate health statistics that accurately reflect the non-institutionalized civilian population throughout the United States. Inclusion criteria for study participants comprised: (1) individuals aged 20 years or older; (2) accessible fasting blood specimens with comprehensive laboratory measurements for high-density lipoprotein (HDL) cholesterol and TG levels; (3) complete body measurement records encompassing BMI and WC; (4) verified self-reported arthritis diagnosis data, particularly distinguishing between RA, osteoarthritis, and other arthritis types; and (5) fully documented covariate information including demographic factors (age, sex, racial/ethnic background, marital status, educational attainment, poverty-income ratio [PIR]), medical conditions (hypertension, diabetes), and lifestyle habits (tobacco use, alcohol intake). Exclusion criteria involved incomplete data necessary for calculating LAP or VAI indices, absence of RA status documentation, or missing values for critical control variables. Assessment of RA The identification of RA relied on self-reported data from the medical conditions questionnaire (MCQ) conducted during NHANES [13]. Participants were required to disclose if any healthcare provider had previously diagnosed them with arthritis, with distinct response categories for RA, osteoarthritis, or alternative forms. Individuals confirming an RA diagnosis were categorized as cases, whereas those denying arthritis or reporting other arthritic conditions were labeled as non-RA subjects. Assessment of LAP and VAI The calculation methods for LAP and VAI were based on sex-specific equations established in prior studies [14]. LAP was computed by combining waist circumference (WC, measured in centimeters) and triglyceride (TG) concentrations (in mmol/L) through different mathematical expressions for each sex: the formula for males was (WC − 65) multiplied by TG, whereas females used (WC − 58) multiplied by TG. VAI determination required supplementary variables such as body mass index (BMI, in kg/m²) and HDL cholesterol levels (mmol/L), with distinct computational approaches for males ([WC divided by (39.68 plus 1.88 multiplied by BMI)] multiplied by [TG divided by 1.03] multiplied by [1.31 divided by HDL]) and females ([WC divided by (39.58 plus 1.89 multiplied by BMI)] multiplied by [TG divided by 0.83] multiplied by [1.52 divided by HDL]). Measurement of covariates The following personal attributes were documented: chronological age (measured as a continuous variable), biological sex (male or female), racial/ethnic background (Mexican American, other Hispanic groups, non-Hispanic White, non-Hispanic Black, and other racial categories including mixed-race individuals), relationship status (partnered or cohabiting versus single), educational attainment (below secondary education, secondary education completion, or post-secondary education), and PIR as a continuous measure. Medical history included self-reported diagnoses of hypertension and diabetes mellitus (with diabetes categorized as confirmed, negative, or borderline). Lifestyle factors incorporated tobacco use history (having smoked a minimum of 100 cigarettes: yes or no) and habitual alcohol intake (regular drinker: yes or no). Anthropometric data included BMI, derived from weight in kilograms divided by height in meters squared, and WC was assessed at the anatomical midpoint between the lower rib margin and the iliac crest. Statistical analysis The statistical analyses were conducted within the R computational platform (release 4.5.0). Considering NHANES’s complex multistage probability-based sampling architecture, appropriate survey weighting parameters were consistently incorporated across all analytical processes to derive nationally representative estimates. The methodological approach leveraged R’s survey package, implementing the svydesign() functionality with calibrated sampling weights that accounted for differential selection probabilities and hierarchical cluster sampling patterns across various data collection cycles. Quantitative outcomes were expressed as weighted means ± standard deviation (SD), while qualitative variables were documented as weighted percentages. Group comparisons (RA vs. non-RA) employed weighted chi-square tests for categorical measures. For continuous parameters with normal distribution, weighted t-tests were conducted, whereas non-normally distributed variables were analyzed through weighted rank-sum tests. Survey-weighted multivariable logistic regression assessed associations between LAP, VAI, and RA, incorporating three sequential adjustment models: Model 1 (unadjusted), Model 2 (demographic covariates: age, sex, race/ethnicity), and Model 3 (fully adjusted for age, sex, race/ethnicity, marital status, education, smoking, alcohol consumption, PIR, hypertension, and diabetes). Both continuous (per-unit increase) and tertile-stratified analyses of LAP and VAI were conducted to investigate potential dose–response relationships. Standardized Z-scores (Z = [X - X̄]/s, where X̄ = mean, s = SD) facilitated direct index comparisons. Restricted cubic spline models examined nonlinear associations between adiposity indices and RA. Upon detecting a threshold effect for LAP in spline analysis, piecewise linear regression identified inflection points and evaluated distinct associations across LAP intervals. Further subgroup analyses were performed to assess potential effect modification across diverse demographic and clinical categories, including age groups (< 60 versus ≥ 60 years), sex, racial/ethnic background, education level, marital status, diabetes status, hypertension status, alcohol consumption, and smoking status. In these stratified assessments, LAP was evaluated using tertile classifications due to its significant association with RA and nonlinear relationship, while VAI was analyzed as a continuous measure given its non-significant association in the fully adjusted model and linear relationship identified in the RCS analysis. All hypothesis testing employed two-tailed approaches, with P < 0.05 considered statistically significant. Results Baseline characteristics Figure 1 illustrates the participant selection process. Initial exclusions involved 46,235 individuals under 20 years old, 34,594 with incomplete LAP, VAI, or RA data, 2,937 lacking covariate details, and 1,632 diagnosed with other arthritic conditions. The final cohort consisted of 15,918 subjects, including 1,988 RA patients and 13,930 non-RA controls. All subsequent analyses employed weighted calculations to maintain population-level accuracy. Flow chart for inclusion and exclusion of study participants Demographic comparisons revealed significant differences between groups. RA patients demonstrated higher median age (60.00 versus 41.00 years, P < 0.001), greater female representation (60.3% compared to 48.3%, P < 0.001), and lower educational attainment (58.1% versus 61.7%, P = 0.03). The RA population exhibited substantially worse health profiles, including increased smoking prevalence (57.1% versus 43.8%, P < 0.001), hypertension incidence (54.6% versus 24.1%, P < 0.001), and diabetes occurrence (17.0% versus 6.0%, P < 0.001). Metabolic measurements showed distinct patterns between cohorts. The RA group displayed elevated median values for BMI (29.58 versus 27.20, P < 0.001), WC (103.20 versus 95.50 cm, P < 0.001), and TG (1.29 versus 1.14, P < 0.001), while HDL levels remained similar (1.34 versus 1.32, P = 0.028). Both LAP (53.93 versus 38.88, P < 0.001) and VAI (1.65 versus 1.39, P < 0.001) scores were significantly higher among RA participants. The proportion of RA cases in the upper tertile was markedly greater for both indices (LAP: 43.7% versus 29.9%; VAI: 38.6% versus 31.1%, both P < 0.001). Table 1 provides comprehensive baseline data using both weighted and unweighted approaches, with weighted results prioritized for national population representation. Table 1Baseline characteristicsUnweightedWeightedTotal (N = 15918)non-RA (N = 13930)RA (N = 1988)P valueTotal (N = 140266008)non-RA (N = 124022032.8)RA (N = 16243975.2)P valueSex (%) Male8027 (49.8)7221 (51.8)806 (40.5)< 0.00170550713.4 (50.3)64103512.3 (51.7)6447201.1 (39.7)< 0.001 Female7891 (50.2)6709 (48.2)1182 (59.5)69715294.6 (49.7)59918520.5 (48.3)9796774.1 (60.3)Age (median [IQR])45.00 [32.00–60.00]43.00 [31.00–58.00]62.50 [52.00–72.00]< 0.00142.00 [30.00–55.00]41.00 [30.00–53.00]60.00 [50.00–69.00]< 0.001Ethnicity (%)< 0.001 Non-Hispanic White2926 (18.4)2683 (19.3)243 (12.2)< 0.00111972773.2 (8.5)11222018.2 (9.0)750755.0 (4.6) Non-Hispanic Black1291 (8.1)1141 (8.2)150 (7.5)7508366.1 (5.4)6967142.0 (5.6)541224.1 (3.3) Mexican American7106 (44.6)6079 (43.6)1027 (51.7)96382776.6 (68.7)84079009.5 (67.8)12303767.1 (75.7) Other Race3124 (19.6)2691 (19.3)433 (21.8)15,111,298 (10.8)13408048.2 (10.8)1703249.8 (10.5) Other Hispanic1471 (9.2)1336 (9.6)135 (6.8)9290804.1 (6.6)8345814.9 (6.7)944979.2 (5.8)Education (%) Low3881 (24.4)3340 (24.0)541 (27.2)0.00221330860.9 (15.2)18416630.9 (14.8)2914230.0 (17.9)0.03 Middle3601 (22.6)3142 (22.5)459 (23.1)32932980.9 (23.5)29044931.1 (23.4)3888049.8 (23.9) High8436 (53.0)7448 (53.5)988 (49.7)86002166.2 (61.3)76560470.7 (61.7)9441695.5 (58.1)Marital Status (%) Married/Living with partner9896 (62.2)8715 (62.6)1181 (59.4)0.00791148354.8 (65.0)80314195.7 (64.8)10834159.1 (66.7)0.241 Non-married6022 (37.8)5215 (37.4)807 (40.6)49117653.2 (35.0)43707837.1 (35.2)5409816.1 (33.3)PIR (median [IQR])2.24 [1.17–4.21]2.27 [1.18–4.25]1.94 [1.09–3.95]< 0.0013.05 [1.55–4.98]3.08 [1.57-5.00]2.72 [1.37–4.93]0.052Alcohol (%) Yes11,798 (74.1)10,363 (74.4)1435 (72.2)0.038110774868.2 (79.0)98263495.7 (79.2)12511372.5 (77.0)0.094 No4120 (25.9)3567 (25.6)553 (27.8)29491139.8 (21.0)25758537.1 (20.8)3732602.7 (23.0)Smoking (%) Yes7176 (45.1)6069 (43.6)1107 (55.7)< 0.00163618896.9 (45.4)54336539.6 (43.8)9282557.3 (57.1)< 0.001 No8742 (54.9)7861 (56.4)881 (44.3)76646911.1 (54.6)69685493.2 (56.2)6961417.9 (42.9)Hypertension (%) Yes4974 (31.3)3800 (27.3)1174 (59.1)< 0.00138772574.5 (27.6)29,905,798 (24.1)8866776.5 (54.6)< 0.001 No10,944 (68.7)10,130 (72.7)814 (40.9)101493433.5 (72.4)94116234.8 (75.9)7377198.7 (45.4)Diabetes (%) Yes1661 (10.4)1207 (8.7)454 (22.8)< 0.00110256979.9 (7.3)7490773.6 (6.0)2766206.3 (17.0)< 0.001 No13,963 (87.7)12,501 (89.7)1462 (73.5)127676567.5 (91.0)114714562.4 (92.5)12962005.1 (79.8) Borderline294 (1.9)222 (1.6)72 (3.6)2332460.6 (1.7)1816696.8 (1.5)515763.8 (3.2)BMI (median [IQR])27.60 [24.05–31.80]27.37 [23.94–31.50]29.60 [25.60–35.00]< 0.00127.30 [23.75–31.45]27.20 [23.69–31.40]29.58 [25.74–34.60]< 0.001WC (median [IQR])97.00 [87.00-107.50]96.00 [86.20-106.50]103.00 [93.70-114.23]< 0.00196.00 [85.60-106.80]95.50 [85.40-106.50]103.20 [93.80-113.80]< 0.001HDL (median [IQR])1.32 [1.09–1.60]1.32 [1.09–1.60]1.34 [1.09–1.66]0.0251.32 [1.09–1.60]1.32 [1.09–1.60]1.34 [1.09–1.66]0.028TG (median [IQR])1.20 [0.82–1.77]1.19 [0.81–1.76]1.28 [0.88–1.83]< 0.0011.15 [0.80–1.70]1.14 [0.79–1.69]1.29 [0.88–1.86]< 0.001LAP (median [IQR])42.50 [23.50–74.00]40.89 [22.06–72.13]54.15 [32.17–88.24]< 0.00140.00 [21.50–72.00]38.88 [20.86–69.70]53.93 [30.64–89.90]< 0.001LAP Tertiles (%) <28.975306 (33.3)4883 (35.1)423 (21.3)< 0.00150040260.4 (35.7)46341887.7 (37.4)3698372.7 (22.8)< 0.001 28.97–60.525306 (33.3)4612 (33.1)694 (34.9)45995693.7 (32.8)40545504.9 (32.7)5450188.8 (33.6) ≥60.525306 (33.3)4435 (31.8)871 (43.8)44230053.8 (31.5)37134640.1 (29.9)7095413.7 (43.7)VAI (median [IQR])1.48 [0.90–2.50]1.45 [0.88–2.45]1.66 [0.99–2.72]< 0.0011.42 [0.87–2.40]1.39 [0.85–2.36]1.65 [0.99–2.71]< 0.001VAI Tertiles (%) <1.065306 (33.3)4740 (34.0)566 (28.5)< 0.00148792610.2 (34.8)44056262.0 (35.5)4736348.2 (29.2)< 0.001 1.06–2.065306 (33.3)4637 (33.3)669 (33.7)46636831.8 (33.3)41405987.4 (33.4)5230844.4 (32.2) ≥2.065306 (33.3)4553 (32.7)753 (37.9)44836566.0 (32.0)38559783.4 (31.1)6276782.6 (38.6)RA Rheumatoid arthritis, BMI Body Mass Index, WC Waist Circumference, HDL High-Density Lipoprotein, TG Triglyceride, VAI Visceral adiposity index, LAP Lipid accumulation product Baseline characteristics RA Rheumatoid arthritis, BMI Body Mass Index, WC Waist Circumference, HDL High-Density Lipoprotein, TG Triglyceride, VAI Visceral adiposity index, LAP Lipid accumulation product Association between LAP, VAI, and RA Table 2 presents the weighted multivariate logistic regression outcomes assessing connections among LAP, VAI, and RA. The crude analysis (Model 1) indicated significant associations for both indices. For continuous LAP measurements, each unit rise was associated with a 0.3% higher RA prevalence (odds ratio [OR] = 1.003, 95% confidence interval [CI]: 1.002–1.004, P < 0.001). When categorized into tertiles, subjects in the middle tertile (Q2) showed 68.4% elevated RA prevalence compared to the lowest tertile (Q1) (OR = 1.684, 95% CI: 1.384–2.050, P < 0.001), with the highest tertile (Q3) demonstrating 139.4% greater prevalence (OR = 2.394, 95% CI: 1.200–2.873, P < 0.001). In Model 1’s continuous VAI analysis, every unit increase was linked to 3.7% higher RA occurrence (OR = 1.037, 95% CI: 1.016–1.059, P < 0.001). Tertile stratification revealed Q2 participants had 17.5% increased prevalence versus Q1 (OR = 1.175, 95% CI: 0.969–1.425, P = 0.100), while Q3 individuals showed 51.4% greater prevalence (OR = 1.514, 95% CI: 1.254–1.828, P < 0.001). Table 2Multivariable logistic regression models for the association between LAP, VAI, and RAModel1Model2Model3OR (95%CI) P OR (95%CI) P OR (95%CI) P LAP per unit1.003 (1.002, 1.004)< 0.0011.003 (1.001, 1.004)< 0.0011.002 (1.001, 1.003)< 0.001 per SD1.229 (1.130, 1.336)< 0.0011.210 (1.100, 1.330)< 0.0011.125 (1.054, 1.200)< 0.001 Z-score standardized1.007 (1.005, 1.009)< 0.0011.007 (1.005, 1.009)< 0.0011.005 (1.003, 1.006)< 0.001 Q1 (< 28.97)Ref.Ref.Ref. Q2 (28.97–60.52)1.684 (1.384, 2.050)< 0.0011.228 (1.002, 1.504)0.0481.114 (0.908, 1.367)0.299 Q3 (≥ 60.52)2.394 (1.200, 2.873)< 0.0011.760 (1.456, 2.128)< 0.0011.410 (1.163, 1.709)< 0.001 P for trend< 0.001< 0.001< 0.001VAI per unit1.037 (1.016, 1.059)< 0.0011.037 (1.014, 1.061)0.0021.019 (0.998, 1.041)0.068 Z-score standardized1.074 (1.034, 1.116)< 0.0011.083 (1.038, 1.130)< 0.0011.043 (0.999, 1.090)0.058 Q1 (< 1.06)Ref.Ref.Ref. Q2 (1.06–2.06)1.175 (0.969, 1.425)0.1001.032 (0.849, 1.254)0.7520.937 (0.765, 1.148)0.526 Q3 (≥ 2.06)1.514 (1.254, 1.828)< 0.0011.276 (1.051, 1.550)0.0141.051 (0.850, 1.265)0.620 P for trend< 0.0010.0140.679Model 1: unadjustedModel 2: adjusted for age, sex, and raceModel 3: adjusted for age, sex, race, marital status, education level, smoking status, alcohol drinking status, poverty income ratio, hypertension, and diabetesRA Rheumatoid arthritis, VAI Visceral adiposity index, LAP Lipid accumulation product, NHANES National Health and Nutrition Examination Survey, OR Odds ratio, CI Confidence interval Multivariable logistic regression models for the association between LAP, VAI, and RA Model 1: unadjusted Model 2: adjusted for age, sex, and race Model 3: adjusted for age, sex, race, marital status, education level, smoking status, alcohol drinking status, poverty income ratio, hypertension, and diabetes RA Rheumatoid arthritis, VAI Visceral adiposity index, LAP Lipid accumulation product, NHANES National Health and Nutrition Examination Survey, OR Odds ratio, CI Confidence interval When adjusting for demographic characteristics in Model 2 (incorporating age, sex, and race/ethnicity), the identified associations retained statistical significance while demonstrating reduced effect magnitudes. For LAP measurements, the highest tertile (Q3) persistently displayed a 76.0% greater occurrence (OR = 1.760, 95% CI: 1.456–2.128, P < 0.001), with a statistically significant progression through tertile divisions (P for trend < 0.001). Concerning VAI, the top tertile manifested a 27.6% rise in prevalence (OR = 1.276, 95% CI: 1.051–1.550, P = 0.014), with an observable dose–response relationship (P for trend = 0.014). The comprehensive Model 3, which accounted for all potential confounding factors, revealed a persistent statistically meaningful association between LAP and RA occurrence. Analyzing LAP as a continuous variable demonstrated that each additional unit was associated with a 0.2% rise in occurrence rate (OR = 1.002, 95% CI: 1.001–1.003, P < 0.001). When evaluating by tertiles, participants in the middle tertile exhibited an 11.4% higher occurrence rate compared to the lowest tertile (OR = 1.114, 95% CI: 0.908–1.367, P = 0.299), while the highest tertile displayed a substantial 41.0% increase (OR = 1.410, 95% CI: 1.163–1.709, P < 0.001), confirming a distinct progressive relationship (P for trend < 0.001). The analysis of VAI measurements revealed attenuated associations after thorough statistical adjustments. When examined as a continuous variable, every incremental unit was linked to a modest 1.9% rise in occurrence rates (OR = 1.019, 95% CI: 0.998–1.041, P = 0.068). In stratified evaluations by tertiles, the middle tertile showed a 6.3% lower prevalence compared to the lowest tertile (OR = 0.937, 95% CI: 0.765–1.148, P = 0.526), whereas the highest tertile demonstrated a slight 5.1% elevation (OR = 1.051, 95% CI: 0.850–1.265, P = 0.620). None of these relationships achieved statistical significance, and the overall pattern assessment yielded non-significant results (P for trend = 0.679). To enable direct comparison between LAP and VAI, Z-score standardization was implemented. Post-standardization analyses revealed that every standard deviation increase in LAP was associated with a 12.5% elevation in RA occurrence rate within the fully adjusted model (OR = 1.125, 95% CI: 1.054–1.200, P < 0.001). In contrast, the standardized VAI measurement exhibited a weaker and statistically non-significant association (OR = 1.043, 95% CI: 0.999–1.090, P = 0.058), further supporting that LAP may perform better than VAI in assessing the association with RA prevalence. Nonlinear associations between LAP, VAI, and RA RCS analysis uncovered divergent association patterns between LAP, VAI, and RA prevalence. Regarding LAP (Fig. 2A), a statistically meaningful overall association was identified (P for overall < 0.001) accompanied by clear nonlinear characteristics (P for nonlinearity < 0.001). The curve exhibited a sharp upward trajectory at initial LAP levels that progressively stabilized with increasing values. Conversely, VAI (Fig. 2B) showed an insignificant overall association (P for overall = 0.397) with no detectable nonlinear pattern (P for nonlinearity = 0.864), suggesting absence of meaningful or nonlinear connections between VAI and RA. The nonlinear associations between the LAP, VAI and RA. A Nonlinear associations between the LAP and RA. B Nonlinear associations between the VAI and RA. RA: Rheumatoid arthritis; VAI: Visceral adiposity index; LAP: Lipid accumulation product Threshold effect analysis Piecewise linear regression was employed to further characterize the nonlinear association between LAP and RA (Table 3). The analysis revealed a transition point occurring at LAP = 43.45. When LAP measurements fell below this threshold (LAP < 43.45), every incremental rise in LAP was associated with a 0.28% elevation in RA prevalence (OR = 1.0028, 95% CI: 1.0022 to 1.0035, P < 0.001). Conversely, for LAP levels surpassing this boundary (LAP > 43.45), the relationship weakened considerably and lost statistical significance (OR = 1.0001, 95% CI: 0.9999 to 1.0002, P = 0.096), indicating a leveling-off phenomenon at elevated LAP concentrations. The log likelihood ratio test substantiated the existence of this threshold effect (P < 0.001). Table 3Threshold effect analysis of LAP on RA using piecewise linear regressionLAP rangeβ (95% CI)P value< 43.451.0028 (1.0022–1.0035)< 0.001≥ 43.451.0001 (0.9999–1.0002)0.096Log likelihood ratio test< 0.001 Threshold effect analysis of LAP on RA using piecewise linear regression Subgroup analyses Subgroup analyses examined the relationship between LAP and RA across diverse demographic categories. Age demonstrated potential modifying effects, with more pronounced correlations detected in younger individuals. For subjects under 60 years, both intermediate and highest LAP tertiles exhibited statistically meaningful links with RA (second tertile vs. lowest: OR = 1.367, 95% CI: 1.027–1.821, P = 0.033; highest tertile vs. lowest: OR = 1.807, 95% CI: 1.405–2.324, P < 0.001). Conversely, these relationships were absent in participants aged 60 or older. Racial/ethnic variations revealed substantial differences, with particularly strong connections in non-Hispanic White and non-Hispanic Black populations. Non-Hispanic White individuals displayed significant associations for both middle and upper LAP tertiles (second tertile vs. lowest: OR = 2.164, 95% CI: 1.211–3.865, P = 0.010; highest tertile vs. lowest: OR = 2.197, 95% CI: 1.198–4.028, P = 0.012). Comparable trends emerged among non-Hispanic Black participants, whereas other racial/ethnic groups showed weaker or insignificant results. Educational background also influenced the LAP-RA relationship, with significant findings concentrated among those with intermediate and higher education (Table 4). Continuous LAP assessment in subgroup evaluations (Fig. 3) yielded similar trends. Meaningful associations appeared in subjects below 60 (OR = 1.002, 95% CI: 1.001–1.030, P = 0.001), both sexes (males: OR = 1.001, 95% CI: 1.000–1.003, P = 0.011; females: OR = 1.002, 95% CI: 1.000–1.005, P = 0.037), and various racial/ethnic categories. Table 4Subgroup analysis of the correlation between LAP and RAQ1 vs. Q2Q1 vs. Q3OR (95%CI) P OR (95%CI) P Age < 601.367 (1.027, 1.821)0.0331.807 (1.405, 2.324)< 0.001 ≥ 600.977 (0.705, 1.355)0.8891.083 (0.812, 1.444)0.585Sex male1.096 (0.812, 1.479)0.5451.351 (1.032, 1.770)0.029 female1.072 (0.818, 1.405)0.6131.372 (1.045, 1.803)0.023Ethnicity Non-Hispanic White2.164 (1.211, 3.865)0.0102.197 (1.198, 4.028)0.012 non-Hispanic Black2.597 (1.213, 5.562)0.0153.524 (1.429, 8.691)0.007 Mexican American1.052 (0.804, 1.377)0.7111.308 (1.033, 1.656)0.026 Other Race1.340 (0.986, 1.821)0.0611.590 (1.139, 2.221)0.007 Other Hispanic0.538 (0.259, 1.114)0.0941.347 (0.632, 2.869)0.436Education Low0.895 (0.578, 1.386)0.6171.067 (0.736, 1.546)0.731 Middle1.154 (0.727, 1.832)0.5411.714 (1.143, 2.570)0.010 High1.148 (0.919, 1.435)0.2221.383 (1.062, 1.762)0.013Marital Married/ living with partner1.015 (0.785, 1.313)0.9081.391 (1.114, 1.736)0.004 Non-married1.299 (0.968, 1.744)0.0811.435 (1.064, 1.935)0.018Diabetes Yes1.400 (0.783, 2.504)0.2542.208 (1.210, 4.030)0.010 No1.096 (0.881, 1.364)0.4071.311 (1.077, 1.596)0.007 Borderline1.084 (0.256, 4.583)0.9101.502 (0.398, 5.661)0.534Hypertension Yes1.336 (0.989, 1.805)0.0591.602 (1.207, 2.126)0.001 No0.934 (0.707, 1.234)0.6291.227 (0.928, 1.622)0.149Alcohol drinking Yes1.058 (0.834, 1.342)0.6381.317 (1.068, 1.625)0.010 No1.375 (0.924, 2.046)0.1161.803 (1.178, 2.759)0.007Smoking Yes0.862 (0.643, 1.155)0.3171.172 (0.914, 1.503)0.208 No1.507 (1.170 1.941)0.0021.784 (1.309, 2.430)< 0.001Adjusted for age, sex, race, marital status, education level, smoking status, alcohol drinking status, poverty income ratio, hypertension, and diabetes Subgroup analysis of the correlation between LAP and RA Adjusted for age, sex, race, marital status, education level, smoking status, alcohol drinking status, poverty income ratio, hypertension, and diabetes Stratified analysis of LAP (continuous variable) with RA across subgroups. RA: Rheumatoid arthritis; LAP: Lipid accumulation product Interaction analyses identified hypertension as a significant modifier (P for interaction = 0.040), demonstrating stronger effects in hypertensive individuals (OR = 1.003, 95% CI: 1.001–1.004, P < 0.001) versus non-hypertensive counterparts (OR = 1.001, 95% CI: 1.000–1.002, P = 0.025). Age showed borderline modification (P for interaction = 0.120), while no significant interactions emerged for sex (P for interaction = 0.200), race/ethnicity (P for interaction = 0.227), education (P for interaction = 0.445), marital status (P for interaction = 0.375), diabetes (P for interaction = 0.280), alcohol consumption (P for interaction = 0.756), or smoking status (P for interaction = 0.565). Concerning VAI, the stratified analyses produced limited statistically meaningful results among different population segments. The most consistent pattern emerged in age-based comparisons, where notable associations were found solely in individuals under 60 years (OR = 1.030, 95% CI: 1.006–1.055, P = 0.016). Within other categories, only alcohol consumers demonstrated a marginally significant relationship (OR = 1.028, 95% CI: 1.002–1.054, P = 0.032). No statistically significant associations were identified between VAI and RA prevalence across the remaining demographic and clinical categories (Table 5). Table 5Subgroup analysis of the correlation between VAI and RASubgroupOR (95%CI) P Age < 601.030 (1.006, 1.055)0.016 >=600.955 (0.906, 1.008)0.092Sex male1.016 (0.980, 1.054)0.390 female1.018 (0.994, 1.042)0.142Ethnicity Non-Hispanic White1.010 (0.931, 1.095)0.816 non-Hispanic Black0.994 (0.960, 1.030)0.753 Mexican American1.026 (0.998, 1.053)0.066 Other Race0.963 (0.883, 1.049)0.383 Other Hispanic0.955 (0.829, 1.100)0.522Education Low1.016 (0.995, 1.038)0.140 Middle1.018 (0.977, 1.061)0.388 High1.019 (0.984, 1.056)0.280Marital Married/ living with partner1.021 (0.991, 1.052)0.171 Non-married1.018 (0.995, 1.042)0.116Diabetes Yes1.010 (0.986, 1.035)0.411 No1.015 (0.987, 1.044)0.283 Borderline1.089 (0.978, 1.214)0.059Hypertension Yes1.028 (0.992, 1.065)0.123 No1.008 (0.971, 1.046)0.680Alcohol drinking Yes1.028 (1.002, 1.054)0.032 No0.971 (0.904, 1.042)0.407Smoking Yes1.014 (0.986, 1.043)0.332 No1.023 (0.993, 1.054)0.130Adjusted for age, sex, race, marital status, education level, smoking status, alcohol drinking status, poverty income ratio, hypertension, and diabetes Subgroup analysis of the correlation between VAI and RA Adjusted for age, sex, race, marital status, education level, smoking status, alcohol drinking status, poverty income ratio, hypertension, and diabetes Discussion This cross-sectional study revealed that LAP retained a statistically significant independent association with RA prevalence following comprehensive adjustment for confounding variables, whereas VAI’s association became non-significant in the fully adjusted model. These findings offer potentially useful perspectives regarding visceral fat accumulation and lipid accumulation in RA pathogenesis, and may support LAP’s potential clinical application as a screening tool for identifying individuals with higher RA prevalence. The positive relationship between LAP and RA aligns with accumulating data on the pro-inflammatory properties of adipose tissue in autoimmune disorders [14]. Modern research classifies adipose tissue as a dynamic endocrine organ that secretes numerous inflammatory mediators such as TNF-α, IL-6, and MCP-1, all contributing to persistent low-level systemic inflammation [15]. Studies demonstrate that visceral adipose tissue displays significantly higher inflammatory activity compared to subcutaneous fat, making it a key driver of metabolic dysfunction and inflammatory disease processes [16]. Regarding RA pathogenesis, excessive visceral fat accumulation may drive disease development through various interrelated biological pathways. Pro-inflammatory cytokines released by dysfunctional adipocytes have been observed to stimulate Th1and Th17 cell differentiation while suppressing regulatory T cell function – processes central to RA development [17]. Additionally, visceral fat accumulation is associated with increased oxidative stress, which can cause lipid peroxidation and cellular damage, further amplifying inflammatory responses and potentially triggering autoimmune processes [18]. The potentially better performance of LAP relative to VAI in the present study may reflect specific metabolic pathways involved in RA pathogenesis. LAP combines waist circumference and triglycerides, two parameters closely related to hepatic insulin resistance and ectopic fat deposition [10]. Notably, the piecewise regression analysis identified a threshold effect at LAP = 43.45, below which the association with RA was significantly stronger, while above this point the relationship plateaued. This threshold effect suggests that visceral fat accumulation may exert its greatest impact on RA prevalence within a specific metabolic range, beyond which compensatory mechanisms or ceiling effects may attenuate further increases. Recent lipidomics studies of RA patients have revealed significant alterations in lipid metabolism, with elevated triglyceride levels and disrupted fatty acid composition being associated with disease activity and inflammatory burden [19]. These metabolic disturbances may represent early pathogenic events preceding clinical RA manifestations, which may partially explain why LAP demonstrated stronger associations with RA prevalence in the present analysis. The present findings are consistent with recent meta-analyses demonstrating positive correlations between obesity indices and RA prevalence, particularly in women. A 2020 systematic review and meta-analysis of cohort studies found that each 5 kg/m² increase in BMI was associated with an 11% increase in RA risk in women, while no association was observed in men [20]. Similarly, the subgroup analyses in this study revealed stronger LAP associations in specific population subgroups, including younger participants and certain racial/ethnic groups, suggesting that metabolic factors may play differential roles across different demographic characteristics. A study using computed tomography to assess abdominal fat distribution found that RA patients had significantly higher visceral fat area compared to participants without RA, despite similar BMI and WC measurements [21]. These findings are consistent with the present results, showing elevated LAP and VAI values in RA participants even after adjusting for traditional anthropometric indices. Moreover, visceral fat accumulation has been associated with increased inflammatory markers and cardiometabolic risk factors in RA populations, suggesting that the metabolic consequences of visceral fat accumulation may be amplified in the context of chronic systemic inflammation [22]. The clinical relevance of the present findings may extend beyond diagnostic applications. Recent research suggests that novel obesity indices like LAP may serve as valuable biomarkers for predicting disease outcomes and treatment response in inflammatory diseases [23]. For example, a prospective study of early RA patients found that higher visceral fat accumulation indices were associated with greater bone loss and inflammatory burden [24]. These observations suggest that LAP assessment may inform clinical decision-making for disease monitoring and therapeutic interventions. One notable finding was the age-related pattern of the LAP-RA association, with a borderline significant interaction. Significant relationships were observed only in participants under 60 years of age. This observation has important mechanistic and clinical implications. Age-related changes in adipose tissue distribution and function may explain these differential associations. Younger individuals typically exhibit greater metabolic flexibility and more sensitive inflammatory pathways, potentially making them more susceptible to the adverse effects of visceral fat accumulation [25]. Additionally, hormonal environments change significantly with age, particularly in postmenopausal women, which may alter the relationship between obesity and inflammatory disease prevalence [26]. The age-specific associations observed in the present study are consistent with recent evidence suggesting that an “obesity paradox” may apply to elderly individuals with inflammatory diseases. Some studies have reported that mild overweight or obesity in older adults may have protective effects on certain health outcomes, possibly due to altered inflammatory responses or improved metabolic reserves [27]. However, this phenomenon requires careful interpretation, as the cross-sectional nature of this study cannot determine whether these age-related differences reflect true protective effects or survival bias. Another noteworthy discovery emerged regarding the modifying influence of hypertension (interaction P-value = 0.040). The relationship linking LAP with RA demonstrated greater strength among hypertensive participants relative to normotensive individuals. This pattern aligns with established biological mechanisms, given that both hypertension and visceral obesity involve overlapping disease processes such as persistent subclinical inflammation, impaired vascular function, and insulin resistance [16]. For patients with elevated blood pressure, the inflammatory environment resulting from visceral fat deposition could intensify due to concurrent vascular inflammation and oxidative damage, possibly establishing a compounding mechanism that promotes autoimmune responses and RA development [28]. These results imply that subjects presenting with both hypertension and increased LAP measurements might constitute an exceptionally vulnerable population requiring enhanced surveillance for early RA detection. Future investigations could prioritize several important research directions. Longitudinal cohort analyses assessing the predictive value of LAP and VAI for incident RA development would provide important evidence for causal relationships. Exploring connections between these metabolic indicators and RA disease activity, therapeutic efficacy, and prognosis could refine treatment protocols. Advanced adiposity measurement techniques, including MRI or CT-based visceral fat quantification, might reveal precise fat distribution patterns relevant to RA pathogenesis [29]. Creating composite biomarker models that integrate LAP with inflammatory markers, genetic risk scores, and clinical variables could improve predictive capabilities. These multidimensional strategies may facilitate tailored preventive approaches and timely therapeutic interventions for RA [30]. Strengths and limitations This investigation demonstrates multiple methodological strengths that enhance the reliability and generalizability of its outcomes. First, the utilization of NHANES data provided a substantial, nationally representative sample with uniform measurement techniques and rigorous quality control measures. The application of survey-specific weighting factors enables reasonable extrapolation of these findings to the wider United States adult population. Furthermore, the analytical strategy implemented comprehensive adjustments for numerous potential confounders across demographic, socioeconomic, and clinical categories, thereby reinforcing the internal consistency of the identified associations. Additionally, the combined use of multivariable logistic regression alongside RCS modeling and piecewise linear regression allowed for the identification of nonlinear trends and critical thresholds, offering more detailed observations than traditional linear methods could provide. Another notable strength lies in the simplicity of calculating LAP, which only necessitates waist circumference and standard triglyceride measurements, making it highly feasible for clinical implementation in various healthcare settings [28]. Lastly, the concurrent evaluation of LAP and VAI through Z-score standardization enabled a robust methodological comparison of their respective performance. However, certain limitations warrant consideration. The study’s cross-sectional nature prevents definitive conclusions regarding causality between adiposity indices and rheumatoid arthritis. Longitudinal research is necessary to determine whether increased LAP or VAI values occur before RA development or result from metabolic changes associated with the condition. Additionally, RA status relied on self-reported information rather than clinical confirmation, potentially leading to classification errors; though previous studies have shown reasonable consistency between patient-reported arthritis and clinical documentation within NHANES [31]. The exclusion of 2,937 subjects with missing covariate information might have introduced selection bias, especially if these individuals differed meaningfully from those included in the analysis. Furthermore, the dataset lacked sufficient detail to examine variations by RA subtype or disease duration—factors that could influence how body fat relates to systemic inflammation. Geographic limitations also apply, as the research focused solely on United States participants. Caution is advised when applying these findings to populations with distinct genetic predispositions, nutritional patterns, or medical systems. Variations in RA occurrence and body composition worldwide should be accounted for when interpreting these results across different population contexts. Conclusion This investigation revealed a significant and independent association between LAP and RA occurrence, exhibiting a nonlinear dose–response relationship and a critical threshold at LAP = 43.45. In contrast, VAI displayed no meaningful connection following comprehensive statistical adjustments. Requiring merely waist measurement and triglyceride assessment, LAP emerges as an efficient screening parameter for RA susceptibility in general healthcare environments, especially among younger populations and hypertensive patients. The established cutoff value could offer initial guidance for risk categorization. Future prospective research should verify the temporal connection between elevated LAP and RA development, while assessing whether metabolic interventions focused on visceral fat accumulation might reduce RA prevalence. Acknowledgements Not applicable. Abbreviations Lipid Accumulation Product Visceral Adiposity Index Restricted cubic spline Rheumatoid arthritis Body mass index Authors’ contributions Jinye Xu, Yunsheng Xu and Fuxin Wei carried out the studies, participated in collecting data, and drafted the manuscript. Yunsheng Xu performed the statistical analysis and participated in its design. Jinye Xu and Fuxin Wei participated in acquisition, analysis, or interpretation of data and draft the manuscript. All authors read and approved the final manuscript. Funding National Key Research and Development Program of China (2025YFC2511600), Shenzhen Key Laboratory of Bone Tissue Repair and Translational Research (NO. ZDSYS20230626091402006). Data availability All data generated or analysed during this study are included in this published article. Declarations Ethics approval and consent to participate Not applicable. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Yunsheng Xu, Email: xuysh9@mail.sysu.edu.cn. Fuxin Wei, Email: weifuxin@mail.sysu.edu.cn. References Associated Data Data Availability Statement All data generated or analysed during this study are included in this published article.