Association between fatty acids and the risk of impaired glucose tolerance and type 2 diabetes mellitus in American adults: NHANES 2005–2016

Type 2 diabetes mellitus (T2DM) is a growing public health threat driven by genetics, lifestyle, and diet. The role of dietary fat—both total amount and specific fatty acid (FA) subtypes—in the development of impaired glucose tolerance (IGT) and T2DM remains controversial. Prior mechanistic and epidemiologic evidence suggests saturated FAs may promote inflammation and insulin resistance, whereas long-chain unsaturated FAs can enhance incretin secretion, yet clear population-level associations using quantitative dietary FA measures are lacking. This study investigates the associations between dietary FA subtypes and the risks of IGT and T2DM in U.S. adults, and examines potential dose–response patterns.

Existing literature presents mixed findings on dietary fats and T2DM risk. Mechanistic studies suggest saturated FAs, particularly long-chain species like palmitate, may impair insulin signaling, while omega-3 FAs exhibit anti-inflammatory actions and may improve insulin sensitivity. Epidemiologic studies have inconsistently linked total or subtype-specific fat intakes with T2DM risk; some cohorts found no association for total SFA, while others reported inverse associations for MUFA and PUFA with T2DM. Biomarker studies of plasma FAs may not reflect dietary intake accurately, complicating dietary recommendations. The controversy may stem from heterogeneity in FA subtypes, measurement limitations, and confounding by non-dietary factors. This study addresses gaps by quantifying dietary FA subtypes and evaluating their associations with IGT and T2DM in a nationally representative sample using standardized definitions and multivariable adjustment.

Design and data source: Serial cross-sectional analysis using six NHANES cycles (2005–2016), a nationally representative, stratified multistage probability sample of U.S. civilians. Participants: Adults with complete fasting plasma glucose (FPG) and 2-hour oral glucose tolerance test (OGTT) data and key covariates. Exclusions included participants with diagnosed diabetes who may have altered diet under medical guidance to mitigate Neyman bias, and those with missing key data. Final analytic sample: 9,082 participants (NGT=6,892; IGT=1,434; T2DM=765). Outcomes: Diagnosed per WHO criteria: T2DM defined as FPG ≥7.0 mmol/L (126 mg/dL) or 2-hour OGTT ≥11.1 mmol/L (200 mg/dL); IGT defined as FPG <7.0 mmol/L with 2-hour OGTT 7.8–11.1 mmol/L (140–200 mg/dL). Exposure: Dietary FA intakes estimated from two 24-hour recalls (one in-person at MEC, one by telephone 3–10 days later). Intakes quantified using USDA Food and Nutrient Database for Dietary Studies; averages used when two recalls available; single recall used if only one available; excluded if both missing. FA groups and subtypes included: total FA (TFA), total saturated FA (TSFA), monounsaturated FA (MUFA), polyunsaturated FA (PUFA), and specific chain-length/subtype FAs (e.g., 4:0, 6:0, 8:0, 10:0, 12:0, 14:0, 16:0, 18:0; MUFA 16:1, 18:1, 20:1, 22:1; PUFA 18:2, 18:3, 18:4, 20:4; n-3 20:5, 22:5, 22:6). Covariates: Age, sex, race/ethnicity (Mexican American, other Hispanic, non-Hispanic White, non-Hispanic Black, other), education (five levels), poverty-to-income ratio (PIR), BMI, smoking (yes/no), hypertension (yes/no), and physical activity (yes/no), obtained via interview and examination. Statistical analysis: NHANES sampling weights applied. Continuous variables summarized by mean±SD or median (IQR); categorical by percentages. Group comparisons used ANOVA or chi-square tests with FDR correction. Logistic regression modeled odds of IGT and T2DM across: Model 1 (unadjusted), Model 2 (adjusted for age, sex), Model 3 (further adjusted for race, BMI, education, PIR, smoking, hypertension, exercise). FA intakes were analyzed continuously and by quartiles (Q1 reference). Factor analysis with varimax rotation identified uncorrelated FA patterns; factor loadings >0.4 considered significant. Two-sided P<0.05 after Benjamini–Hochberg FDR considered significant. Analyses conducted in R 4.0.2.

- Sample: 9,082 adults (NGT 6,892; IGT 1,434; T2DM 765). IGT and T2DM groups were older, had higher BMI, lower education, less smoking, more hypertension, and less exercise than NGT (all P<0.05). - Mean intakes: TFA, MUFA, and PUFA were significantly lower in IGT and T2DM than in NGT; multiple SFA subtypes also lower in IGT than in T2DM and NGT. - Continuous models (Table 3): - IGT: After full adjustment (Model 3), TSFA inversely associated with IGT (OR 0.991, 95% CI 0.985–0.998, P=0.024). Specific FAs inversely associated with IGT included 8:0 (OR 0.665, 95% CI 0.455–0.972, P=0.035), 10:0 (OR 0.755, 95% CI 0.594–0.959, P=0.025), 14:0 (OR 0.931, 95% CI 0.877–0.987, P=0.024), 18:0 (OR 0.965, 95% CI 0.941–0.991, P=0.024), and 18:3 (OR 0.902, 95% CI 0.831–0.980, P=0.024). - T2DM: No individual FA group or subtype remained significant after full adjustment in continuous analyses. - Quartile analyses: - IGT (Table 4, Model 3): Several SFA subtypes showed lower odds at higher quartiles vs Q1, including 6:0 (Q3 OR 0.747, 95% CI 0.605–0.921, P=0.024), 8:0 (Q4 OR 0.701, 95% CI 0.550–0.894, P=0.021), 10:0 (Q4 OR 0.662, 95% CI 0.510–0.860, P=0.021), and 14:0 (Q4 OR 0.688, 95% CI 0.534–0.887, P=0.021). - T2DM (Table 5): Total FA intake at Q4 vs Q1 was inversely associated with T2DM after full adjustment (OR 0.714, 95% CI 0.532–0.959, P=0.025). Other FA groups/subtypes did not show consistent significant associations after adjustment. - Factor analysis (Table 6): Four FA factors identified. Factor 4 (F4) was inversely associated with T2DM in fully adjusted models (OR 0.824, 95% CI 0.715–0.949, P=0.029). Based on F4 loadings, an unsaturated FA signature was identified comprising 18:1 (octadecenoic), 18:2 (octadecadienoic), 18:3 (octadecatrienoic), and 20:1 (eicosenoic), suggesting these unsaturated FAs may relate to lower T2DM risk. - Overall: Higher TSFA intake was associated with lower odds of IGT, and higher total FA intake (top quartile) with lower odds of T2DM; a pattern of specific unsaturated FAs was linked to lower T2DM risk.

The study addressed whether dietary FA subtypes and their intake levels are associated with IGT and T2DM in U.S. adults using quantitative dietary assessments and robust adjustment for confounders. Findings suggest that higher TSFA intake was associated with lower odds of IGT, and that overall higher total FA intake (Q4 vs Q1) was associated with lower odds of T2DM. While individual MUFA/PUFA groups were not consistently significant after full adjustment, factor analysis revealed a pattern driven by specific unsaturated FAs (18:1, 18:2, 18:3, 20:1) inversely associated with T2DM, implying that focusing on specific FA species rather than broad MUFA/PUFA categories may be more informative. These results align with some prior prospective studies showing inverse associations for unsaturated fats with T2DM risk, while also highlighting inconsistencies likely due to heterogeneity in FA subtypes, dietary assessment limitations, and residual confounding. Mechanistically, unsaturated FAs may exert beneficial effects through anti-inflammatory pathways and incretin-mediated insulin secretion, whereas some SFAs may not uniformly confer risk when examined at specific chain lengths and intake ranges. The study thereby refines understanding of dietary FAs in dysglycemia, suggesting nuanced relationships by FA subtype and intake level.

In a nationally representative U.S. adult sample, higher total saturated fat intake was associated with lower odds of IGT, and higher total fat intake (top quartile) with lower odds of T2DM after multivariable adjustment. A dietary pattern characterized by specific unsaturated FAs (18:1, 18:2, 18:3, 20:1) was inversely associated with T2DM, suggesting that certain unsaturated FAs abundant in natural oils may be protective. Future research should include prospective cohort and interventional studies with improved long-term dietary assessment to establish causality and clarify optimal types and amounts of FA subtypes for diabetes prevention.

- Cross-sectional design precludes causal inference. - Dietary exposure assessed via two 24-hour recalls reflects short-term intake; lack of food frequency questionnaire limits long-term intake estimation. - Potential dietary assessment bias and measurement error. - Residual confounding by non-dietary factors may persist despite adjustment. - Relatively smaller numbers of IGT and T2DM cases compared to NGT may limit power for subtype analyses. - Plasma FA biomarkers were not used; dietary FA might not fully correspond to circulating levels.