Skip to main content
. 2021 Jan 29;113(3):593–601. doi: 10.1093/ajcn/nqaa362

TABLE 3.

Multivariable analyses

Parameter Condition AOR (95% CI)1 P value
Hyperglycemia — 2.48 (1.46–4.20) 0.0008
Hypertriglyceridemia — 2.18 (1.29–3.67) 0.004
MetS2 — 3.20 (1.92–5.32) <0.0001
TyG index2 — 2.57 (1.78–3.71) <0.0001
Alcohol, drinks/week — 1.06 (0.99–1.13) 0.08
BMI, per 5 kg/m2 × Food insecurity (interaction) <0.0001
Food-secure 1.32 (1.04–1.67) 0.02
Food-insecure 3.83 (2.37–6.19) <0.0001
Food insecurity BMI = 25.0 kg/m2 0.20 (0.07–0.57) 0.002
BMI = 30.0 kg/m2 0.59 (0.31–1.14) 0.12
BMI = 35.0 kg/m2 1.91 (1.02–3.59) 0.04
× Obesity, BMI ≥ 30 3 (interaction) 0.003
Nonobese 0.30 (0.11–0.83) 0.02
Obese 1.85 (0.98–3.53) 0.06

Data are for NAFLD (liver fat > 5%), determined by MRI-PDFF (n = 603). Abbreviations: AOR, adjusted odds ratio; MetS, metabolic syndrome; NAFLD, nonalcoholic fatty liver disease; PDFF, proton density fat fraction; TyG, triglyceride and glucose.

1

The multivariable logistic regression model examined the relationship between food insecurity, BMI, and NAFLD, adjusting for age, sex, race/ethnicity, household size, hyperglycemia, and hypertriglyceridemia, and included an interaction term for food insecurity and BMI. Estimates for the effect of food insecurity on NAFLD were obtained for BMIs of 25, 30, and 35 kg/m2, which correspond to the cutoffs for overweight, obesity Class I, and obesity Class II, respectively.

2

Due to multicollinearity, separate models were used to obtain estimates for MetS and TyG, adjusting for age, sex, race/ethnicity, household size, BMI, alcohol consumption, and food insecurity.

3

The multivariable logistic regression model examined the relationship between food insecurity and NAFLD among obese and nonobese participants, adjusting for age, sex, race/ethnicity, household size, hyperglycemia, hypertriglyceridemia, and included obesity (BMI ≥ 30 kg/m2), as well as an interaction term for food insecurity and obesity.