Figure 2:

Incident Type 2 Diabetes Over a Six-Year Period, Stratified by Baseline Dysglycemia

*P<.05 for incident type 2 diabetes when comparing avocado consumer at baseline vs. non-consumers, in cox-proportional hazard models which correct for the complex survey design and control for age, gender, energy intake (kilocalories / day), education level, acculturation, Hispanic/Latino heritage, alcohol intake, physical activity (metabolic equivalent minutes / day), diet quality (alternative Healthy Eating Index [aHEI] score) and smoking status as covariates (fully adjusted model).