Table 1. Estimating informative priors used for slope of CD4 count model.
| Variable | Model Coefficients a | Informative priors | ||||||
|---|---|---|---|---|---|---|---|---|
| Historical study | Current study (Est.) | model 1 (Est.) | Changeb | Coef.c | Variance (Var.) | Precision (1/Var.) | ||
| Est. | Upper CI | |||||||
| Female-sex [36] n = 459 | 35.2 | 66.5 | 36.17 | 24.1 | -13.17 | 23.16 | 117055 | 8.54E-06 |
|
Baseline log10 viral load (copies/mL) [36] n = 459 |
13.9 | 22.2 | -10.09 | 7.5 | -11.89 | 11.31 | 8231.08 | 1.22E-4 |
|
Time on treatment (years) [25] n = 12946 |
65 | 69 | 56.18 | 52.3 | -0.68 | 61.12 | 53919.2 | 1.86E-05 |
|
Time on treatment (years)-squared [25] n = 12946 |
-6 | -5 | -22.97 | -23.0 | -0.27 | -6.03 | 3369.95 | 2.97E-04 |
a list of coefficients from historical simple linear mixed-effects models as published in the respective articles (historical study) and from the same reanalysed models using our data (current study Est.)
b obtained by subtracting current study coefficients from full model (model 1)
c obtained by adding both Change and historical study Est. columns.