Skip to main content

Table 4 Multivariable logistic regression models for association between HIV status and hypertension to assess for confounding

From: Hypertension, kidney disease, HIV and antiretroviral therapy among Tanzanian adults: a cross-sectional study

Model Controls (number = 153) HIV-infected, ART naive P-value versus control HIV-infected, on ART P-value versus control Likelihood ratio test (compared to unadjusted model)
(number = 151) (number = 150)
1) Unadjusted 1 0.28 [0.12 to 0.66] 0.003 2.06 [1.18 to 3.59] 0.01 0
2) Adjusted for age + sex 1 0.32 [0.14 to 0.75] 0.009 2.13 [1.18 to 3.85] 0.01 21.7, 2 d.f.
3) Adjusted for age + sex + body mass index (BMI) 1 0.34 [0.14 to 0.79] 0.01 2.04 [1.12 to 3.67] 0.02 25.7, 3 d.f.
4) Adjusted for age + sex + waist-hip ratio (WHR) 1 0.31 [0.13 to 0.73] 0.007 2.04 [1.12 to 3.71] 0.02 22.2, 3 d.f.
5) Adjusted for age + sex + BMI + vigorous work + alcohol 1 0.35 [0.15 to 0.84] 0.02 2.19 [1.18 to 4.05] 0.01 37.7, 5 d.f. a
6) Adjusted for age + sex + WHR + vigorous work + alcohol 1 0.32 [0.14 to 0.77] 0.01 2.19 [1.17 to 4.10] 0.01 35.4, 5 d.f.
  1. aBest fit model. All models are comparing to HIV-negative controls. Models 1, 2, 3 and 4 were predetermined based on most likely confounders. Models 5 + 6 included other baseline characteristics significantly associated with hypertension in the minimally-adjusted model. ART, antiretroviral therapy.