Skip to main content

Table 4 Covariate-adjusted mediating effects of BMI and WC on the association between HAART and blood pressure

From: Impact of body fat changes in mediating the effects of antiretroviral therapy on blood pressure in HIV-infected persons in a sub-Saharan African setting

Mediator variable

Predictor variable

Outcome variable

Path AB*

% mediated*

Path AB†

% mediated†

Path AB

% mediated

Coefficient (95 % CI)

Coefficient (95 % CI)

Coefficient (95 % CI)

BMI

HAART

Systolic BP

1.32 (0.30 to 2.62)

14

1.24 (0.21 to 2.69)

13

0.85 (−0.67 to 2.38)NS

9

Diastolic BP

0.74 (0.20 to 1.92)

10

0.68 (0.30 to 1.52)

9

0.45 (−0.40 to 1.27)NS

6

WC

HAART

Systolic BP

1.14 (0.21 to 3.19)

13

1.07 (0.16 to 2.74)

12

1.01 (0.33 to 2.52)

11

Diastolic BP

0.76 (0.16 to 1.44)

10

0.71 (0.12 to 1.65)

10

0.68 (0.26 to 1.89)

9

Combined BMI/WC

HAART

Systolic BP

1.40 (0.33 to 3.01)

15

1.40 (0.33 to 3.01)

15

1.00 (−0.40 to 3.63)NS

11

Diastolic BP

0.88 (0.26 to 1.84)

11

0.88 (0.26 to 1.84)

11

0.70 (−0.18 to 2.13)NS

9

  1. Path AB* – the indirect effect of HAART on the outcome variable through the mediator variable, controlled for age (continuous) and sex
  2. Path AB† − the indirect effect of HAART on the outcome variable through the mediator variable, controlled for age (continuous), sex and smoking status
  3. Path AB – the indirect effect of HAART on the outcome variable through the mediator variable, controlled for age (continuous), sex, smoking status, CD4 count (continuous), and HIV duration
  4. Coefficients for paths A, B and C’ are as shown in Table 3
  5. % mediated* – proportion of the total effect of HAART on each outcome variable that is attributed indirectly to the mediator variable, adjusted for age and sex
  6. % mediated† − proportion of the total effect of HAART on each outcome variable that is attributed indirectly to the mediator variable, adjusted for age, sex and smoking
  7. % mediated – proportion of the total effect of the independent variable on the dependent variable that is attributed indirectly to the mediator variable, adjusted for age, sex, smoking status, CD4 cell count and HIV duration
  8. NS – not statistically significant at 5 % level; all other coefficients were statistically significant
  9. 95 % confidence intervals for indirect effects (Path AB) are bias-corrected
  10. Abbreviations: BMI body mass index, BP systolic blood pressure, CI confidence interval, HAART highly active antiretroviral therapy, WC waist circumference