Table 2.
Cross-sectional associations between Normalized Difference Vegetation Index (0.11 unit increase) and components of allostatic load using Multiple Imputationa among Participants from four Sites in Three sub-Saharan African Countries
| Total Population (N=1178) | Urban Population (N=978) | |||||
|---|---|---|---|---|---|---|
| Outcome | β (95% CI) | OR (95% CI) | p | β (95% CI) | OR (95% CI) | p |
| BMI (continuous) | ||||||
| Unadjusted | −1.65 (−1.84, −1.45) | <.0001 | −1.87 (−2.17, −1.58) | <.0001 | ||
| Age + Sex | −1.38 (−1.58, −1.17) | <.0001 | −1.52 (−1.84, −1.21) | <.0001 | ||
| Full Adjustmentb | −1.01 (−1.35, −0.67) | <.0001d | −0.97 (−1.37, −0.57) | <.0001d | ||
| Overweight/Obese | ||||||
| Unadjusted | 0.57 (0.53, 0.62) | <.0001 | 0.52 (0.46, 0.58) | <.0001 | ||
| Age + Sex | 0.64 (0.58, 0.70) | <.0001 | 0.60 (0.53, 0.68) | <.0001 | ||
| Full Adjustmentb | 0.73 (0.62, 0.85) | <.0001d | 0.73 (0.63, 0.85) | <.0001d | ||
| Diabetes | ||||||
| Unadjusted | 0.87 (0.79, 0.97) | 0.0153 | 0.74 (0.63, 0.87) | 0.0003 | ||
| Age + Sex | 0.93 (0.83, 1.05) | 0.2381 | 0.76 (0.64, 0.91) | 0.0032 | ||
| Full Adjustmentb | 0.77 (0.62, 0.96) | 0.0193 | 0.77 (0.62, 0.95) | 0.0167 | ||
| Hypertension | ||||||
| Unadjusted | 0.68 (0.63, 0.74) | <.0001 | 0.75 (0.68, 0.82) | <.0001 | ||
| Age + Sex | 0.76 (0.70, 0.82) | <.0001 | 0.84 (0.75, 0.94) | 0.0017 | ||
| Full Adjustmentb | 0.92 (0.80, 1.05) | 0.1996 | 0.91 (0.79, 1.04) | 0.1531 | ||
| Cholesterol (continuous)c | ||||||
| Unadjusted | −0.24 (−0.30, −0.18) | <.0001 | ||||
| Age + Sex | −0.17 (−0.23, −0.11) | <.0001 | ||||
| Full Adjustmentb | 0.04 (−0.07, 0.14) | 0.4875 | ||||
| Cholesterol (binary)c | ||||||
| Unadjusted | 0.99 (0.90, 1.11) | 0.9252 | ||||
| Age + Sex | 1.07 (0.95, 1.20) | 0.2626 | ||||
| Full Adjustmentb | 1.15 (0.93, 1.42) | 0.2093 | ||||
| Allostatic Load | ||||||
| Unadjusted | ||||||
| 0 (Low) | ||||||
| 1 | 0.86 (0.79, 0.94) | 0.0016 | 0.80 (0.71, 0.91) | 0.0005 | ||
| 2 | 0.66 (0.58, 0.75) | <.0001 | 0.60 (0.50, 0.70) | <.0001 | ||
| ≥ 3(High) | 0.61 (0.52, 0.70) | <.0001 | 0.58 (0.48, 0.69) | <.0001 | ||
| Age + Sex | ||||||
| 0 (Low) | Ref | |||||
| 1 | 0.94 (0.86, 1.04) | 0.1376 | 0.90 (0.78, 1.03) | 0.1376 | ||
| 2 | 0.74 (0.65, 0.84) | <.0001 | 0.66 (0.55, 0.79) | <.0001 | ||
| ≥ 3(High) | 0.69 (0.59, 0.81) | <.0001 | 0.66 (0.54, 0.8) | <.0001 | ||
| Full Adjustmentb | ||||||
| 0 (Low) | Ref | |||||
| 1 | 0.87 (0.74, 1.02) | 0.1702 | 0.88 (0.74, 1.05) | 0.1702 | ||
| 2 | 0.67 (0.52, 0.85) | 0.0004d | 0.65 (0.51, 0.82) | 0.0004d | ||
| ≥ 3 (High) | 0.66 (0.52, 0.85) | 0.0014d | 0.67 (0.52, 0.85) | 0.0014d | ||
Imputation models were fit using multiple chained equations to impute missing data for body mass index and the following covariates: age, sex, educational attainment, fuel source, marital status, NDVI, smoking status, site.
Multiple regression models were fit for categorical (multinomial logistic), binary (logistic) or continuous (linear) outcome variables, using the following covariates: age, sex, educational attainment, fuel source, marital status, smoking status, site, seeking health care in private sector.
Since cholesterol was missing for one site (peri-urban Uganda), these models were fit only on remaining sites (Tanzania, South Africa, and Rural Uganda) with a sample size of 903.
p<0.05 after Bonferroni correction applied to final fully adjusted outcome models (the p-value cutoff for Bonferroni correction = 0.05/12 outcomes = 0.0042).