Table 4

Odds ratios for baseline and time-varying predictors of adherence ≥ 90%.

Baseline characteristics

Univariate model

OR (95% CI) P*

Multivariate model†

AOR (95% CI) P*


Dosing schedule (prophylaxis group)

Three times a week

1

1

Daily

0.88 (0.66, 1.17) 0.38

0.85 (0.64, 1.11) 0.23


Study site

RCCH

1

1

TCH

1.25 (0.95, 1.67) 0.11

1.12 (0.82, 1.54) 0.46


Gender

Male

1

1

Female

1.17 (0.87, 1.57) 0.30

1.11 (0.82, 1.50) 0.51


Number of individuals in house

≤ 5

1

1

> 5

0.77 (0.57 1.02) 0.07

0.71 (0.54, 0.95) 0.02


Brick house

No

1

-

Yes

0.98 (0.73, 1.30) 0.87


Tap water in house

No

1

-

Yes

1.07 (0.80, 1.43) 0.64


Flush toilet in house

No

1

-

Yes

1.08 (0.82, 1.44) 0.57


Electricity in house

No

1

-

Yes

0.93 (0.61, 1.41) 0.74


Time-varying characteristics


Age at visit

< 1 year

1

1

1-4 years

0.77 (0.6, 0.98) 0.03

1.47 (0.91, 2.37) 0.11

> 4 years

1.40 (1.09, 1.80) 0.008

1.96 (1.16, 3.32) 0.012


Time on study at visit

< 1 year

1

1

≥ 1 year

0.82 (0.67, 1.01) 0.06

0.83 (0.66, 1.05) 0.12


CDC clinical

N or A

1

1

stage at visit

B

1.11 (0.85, 1.46) 0.44

0.85 (0.54, 1.34) 0.48

C

0.86 (0.65, 1.14) 0.30

0.71 (0.43, 1.16) 0.17


CDC immune

1

1

1

Stage at visit

2

0.94 (0.73, 1.20) 0.62

0.96 (0.69, 1.34) 0.83

3

1.02 (0.79, 1.32) 0.86

1.00 (0.70, 1.44) 0.99


Study drug at visit

Placebo

1

-

Isoniazid

1.02 (0.73, 1.43) 0.91

-


On HAART at visit

No

1

1

Yes

1.14 (0.89, 1.45) 0.30

1.08 (0.82, 1.43) 0.56


† Adjusted for group, site, sex, number of people in house, age at visit, concurrent HAART, CDC immune stage at visit, CDC clinical stage at visit, time on study at visit. Two patients excluded from multivariate model due to missing information on number of people in the house and CDC immune staging at follow up.

OR = odds ratio; AOR = adjusted OR; CI = confidence interval; RCCH = Red Cross War Memorial Children's Hospital; TCH = Tygerberg Children's Hospital; CDC = Centers for Disease Control and Prevention; HAART = highly active antiretroviral therapy; * P values from Wald tests

le Roux et al. BMC Medicine 2009 7:67   doi:10.1186/1741-7015-7-67

Open Data