Table 2.

Parameter estimates from weighted regression models examining the association of average avocado intake (servings/day) with average glucose levels (defined by hemoglobin A1c).

Model	Overall Population		Normoglycemic		Prediabetic		Diabetic
	β(SE)	P	β(SE)	P	β(SE)	P	All		Untreated		Treated
							β(SE)	P	β(SE)	P	Metformin		Insulin
											β(SE)	P	β(SE)	P
	N = 14,536a		N = 5,896a		N = 5,993a		N = 3003a		N = 1,495a		N = 1,237a		N = 317a
1	−0.11 (0.05)	.03	−0.004 (0.05)	.94	−0.01 (0.08)	.88	−0.37 (0.18)	.02	−0.35 (0.22)	.11	−0.26 (0.17)	.12	0.07 (0.24)	.78
2	−0.11 (0.05)	.04	0.002 (0.05)	.97	−0.01 (0.08)	.94	−0.37 (0.16)	.02	−0.35 (0.21)	.10	−0.26 (0.17)	.13	0.07 (0.24)	.77
3	−0.08 (0.05)	.15	0.02 (0.06)	.83	0.01 (0.07)	.88	−0.37 (0.16)	.02	−0.36 (0.21)	.08	−0.27 (0.17)	.12	0.07 (0.23)	.76

Note: Significant results (P < 0.05) in bold.

Model 1 includes age, gender, energy intake (kilocalories/day), education level, acculturation, and Hispanic/Latino heritage as covariates (minimally adjusted model).

Model 2 includes the covariates from model 1, with the addition of alcohol intake, physical activity (metabolic equivalent minutes/day), diet quality (alternative Healthy Eating Index [aHEI] score) and smoking status (fully adjusted model).

Model 3 includes the covariates from model 2, with the addition of body mass index (BMI; fully adjusted + BMI model).

^aUnweighted frequency.