Skip to main content
. 2022 Jun 22;116(4):1078–1090. doi: 10.1093/ajcn/nqac174

TABLE 3.

Prospective associations between plant-based diet indices and change in cardiometabolic risk markers among participants with South Asian ancestry, aged 40–84 years, in the MASALA study1

PDI hPDI uPDI
n 2 Model3 Score P value Score P value Score P value
Glycemia measures among nondiabetics
 Fasting glucose,4 mmol/L 556 1 −0.400 ± 0.321 0.21 −0.348 ± 0.271 0.20 0.003 ± 0.322 0.99
2 −0.374 ± 0.320 0.24 −0.313 ± 0.270 0.25 0.015 ± 0.321 0.96
 HbA1c, % 553 1 0.064 ± 0.170 0.71 0.034 ± 0.145 0.82 0.158 ± 0.170 0.36
2 0.064 ± 0.171 0.71 0.033 ± 0.145 0.82 0.158 ± 0.171 0.36
Lipids
 Triglycerides,4 mmol/L 733 1 −0.629 ± 0.964 0.51 −0.499 ± 0.821 0.54 0.598 ± 0.993 0.55
2 −0.608 ± 0.966 0.53 −0.413 ± 0.825 0.61 0.546 ± 0.993 0.58
 HDL-C, mmol/L 732 1 0.005 ± 0.006 0.36 0.003 ± 0.005 0.52 0.004 ± 0.006 0.51
2 0.006 ± 0.006 0.31 0.004 ± 0.005 0.46 0.004 ± 0.006 0.46
 LDL-C, mmol/L 728 1 −0.027 ± 0.021 0.19 −0.006 ± 0.017 0.74 0.005 ± 0.021 0.81
2 −0.026 ± 0.021 0.20 −0.007 ± 0.017 0.70 0.007 ± 0.021 0.74
Body composition measures
 Weight, kg 730 1 −0.209 ± 0.106 0.05 0.004 ± 0.091 0.96 −0.072 ± 0.109 0.51
1

Values represent multivariable-adjusted changes in cardiometabolic risk markers (β ± SE or % increase ± SE for log-transformed variables) for each 5-unit increase in plant-based diet scores, calculated using multivariable linear regression. Abbreviations: HbA1C, glycated hemoglobin; hPDI, healthy plant–based diet index; MASALA, Mediators of Atherosclerosis in South Asians Living in America; PDI, plant-based diet index; uPDI, unhealthy plant–based diet index.

2

Numbers of participants vary due to missing values for outcome variables or to outliers.

3

Multivariable-adjusted model 1 was adjusted for age, sex, study site, education (Bachelor's degree or higher, yes compared with no), smoking status (never, former, current), alcohol (yes compared with no), family history of diabetes (any first-degree biological relatives), years lived in the United States, physical activity (MET-min/week), total energy, diabetes medication use, cholesterol-lowering medication use (yes compared with no), hypertension medication use (yes compared with no), the sum of cultural traditional measures, and the baseline value of the corresponding cardiometabolic risk marker using multivariable linear regression. Multivariable model 2 was additionally adjusted for BMI (kg/m2).

4

Values were log-transformed to approximate a normal distribution of the residuals.